Nuclear Medicine and PET CT Technology and Techniques Nuclear Medicine and PET

ee Ae SOL cs and Statistics |APTER OUTLINE DAMENTALS Scientific Notation "Fractions and Percentages PRACTICAL APPLICATIONS Radioactive Decay qto-""Te Radionuclide Generators “Algebraic Equations and Ratios Half-life: Biological, Physical, and Inverse Square Law Effective “Units Attenuation of Radiation “Exponent Laws and Logarithms Graphs Numeric Accuracy: Significance and Meaturement of Effective Half-Life Rounding Least Squares Curve Fitting “Caleulators and Personal Computers Other Graphs stanisrics Mean and Standard Deviation Gaussian Distributions Poisson Distributions and Counting Statistics Chi-square Tests tests Medical Decision Making half-valne layer (VID Poisson distribution natural logarithin physical half-life Use scientific notation in performing algebraic operations Use the inverse square law to calculate the intensity of a radiation field at various distances + Perform radioactive dilation calculations, + Define the unis of radioactivity, radiation exposure, Tadiation absorbed dose, and radiation dose ‘equivalent. “Perform calculations with logarithms and exponents “ising calculator. Discuss mumeric accuracy, significant digits, and rounding Calculate quantities of radioactivity using the general form of the decay equation and decay factors. ‘Compute the concentration of Mo itt “Te. ‘Compuce effective half-life and biological halflife [uclear medicine technology occupies a unique posi- nn the allied health sciences because of its strong dependence on quantitative, or mathematical, results. This pler attempts to provide a sound basis for performing caléulations that are typically required of nuclear Riicine technologists. The emphasis here is on practical “se, not on theoretical principles. Practical examples are "provided, and each type of calculation includes a discus- hon on the ase of pocket calculators and rudimentary ‘inverse square law proportional Teast squares curve fit scientific notation Iinear attenuation coefficient sensitivity Jogaithm significant figures mean specificity standard deviation (SD) Calculate intensity with half-value layers, Diagram various types of graphs and graphing techniques. Discuss curve-fitting techniques. Define mean, standard deviation, and coefficient of variation Discuss Gaussian and Poisson distributions State the formula for standard deviation, and perform calculations in the presence of background. Explain the function of the chi-square test and interpretation of results. Interpret the results of a chi-square test using @ probability table. Discuss the use and interpretation of tests. Describe the interpretation of sensitivity, specificity, prevalence, and accuracy. Instrnctions for the use of Microsoft Excel. This chapter reviews elementary algebra, graphing techniques, and statistical principles, always with an emphasis on practical applications. I is presumed that readers have a working knowledge of high school-level mathematics. The more advanced reader may wish to skip to the Practical Applica- tions section of this chapter. The reader who requites a more basic review of algebra may wish to consult another mathematics text!2 Nuclear Medicine and PET/CT Technology and Techniques FUNDAMENTALS Scientific Notation Numbers in scientific calculations are often very small, such as 0.0015, or very lange, such as 23,000. Scientific notation allows these numbers to be presemed in a more convenient novation without cumbersome commas and zeros: 1.5 10” and 2.3E x 10°, The exponent on the 10 specifies how many places the decimal point in the number is to be shifted to the left (for negative exponents) or shifted to the right (for positive exponents). Sometimes these ‘numbers are alternatively represented as I.5E-3 and 2.3E+4, a form found on pocket calculators or in printed tabular data (eg, im calculations of internal radiation dositnetty), where the superscript is cumbersome to print Pocket calculators typically have an “EE” key (for “enter exponent’), which allows easy entry of data in scientiic notation, For example, i itis desized to calculate: (15% 10) +3109 the procedure on the calculator would be as follows (Figue 1+) 15 enter value for numerator EE enter exponent key +/— change sign key to make the exponent negative 3 emer the value of the exponent + divide key 23 enter the value for denominator EE enter exponent key 4 enter the value of the exponent = Gee result of 6.5217391E-8 in display) ‘This is precisely the same result as would be obtained by dividing 0.0015 by 23,000 on the calculator. Its often desirable to use numeric prefixes to represent very small or large numbers (Table 1-1). The value 6.52 X 10° grams (g), or 652E-8g, can be more conveniently represented by converting the exponent to one of the exponent values shown in Table 1-1, which are usually exponents divisible by 3 (such as 3, 6, 9, and so on). Thus 6.52 x 10g can be represented as 65.2 X 10°°g, or in convenient shorthand form as 65.2 nanograms (ng), where nano stands for 10°. Notice that the decimal point in the number 6.52 x 10° can be shifted to the right by making the exponent smaller by 1 for each right shift of the decimal (¢ g, 6.52 x 10* = 65.2 x 107), the object being to have an exponent that is divisible by 3 so that the mumeric prefixes in Table T-L can be used. Similarly, a number like 2.3 x 10" counts Gd) can be expressed a5 23.0 x 10?et, or 23 ket, with the k denoting kilo, or thousand. This type of notation is often used for units of radioac- sivity, which is measured in becquetels (Bq). A patient might be injected with 7.4 x 10°Bq of radioactivity, which can be ‘written 0.74 x 10”Bg (the decimal point can be shifted lef if the exponent is increased by 1 for each lelt shift). So the Figuce 11 Generic scientific caleulator demonstrating keys useful in nuclear medicine, The 2nd key produces the function above the key (eg, 2nd) and X, produces the mean of X, values. The INV key produces the functional inverse of the Key. INV and InX are e* functions. Mos caleulators do not have an e* key if they oller INV and InX keys. The FIX key sets the suraber of decimal places to be displaced (e.g, FIX and 3 will display only three decimal places for all subsequent calculations). The ENG ey causes the calculator to display results in scientific notation. The X, key i used to enter x values to calculate the mean ¥ and standerd deviation 6 The X, and Y, Keys are used to enter x and y values for Teast squares curve fit (linear regression) to produce a straight line with intercept a, slope by correlation r, end standard error of estimate SEE. The +/- key changes the sign of 2 numberjed radioactivity could be represented as 0.74 GBQ, Alternately, the value 7-4 x 10*Bq could be written 740.0 x 10" Bq = 740MiBq (the decimal was shifted two places to the tight, sothe exponent is decreased by 2). Choosing between calculators often have an engineering notation key, ich conirols the scientific notation exponents in the (4 x 10°Bq is displayed as 740E6 on the calculator, which user understands to be the same as 740 MBq, tions and Percentages tions, such as, consist of numerator (1) that is to be ded by a denominator (3), The value of a fraction may be expressed decimally, such as Yj = 0.750 (a fraction G@ fraction ipulaion of fractions requires care as to the number of gis and placement ofthe decimal point. The safest way to the mathematical manipulation of fractions is ta lations such as ' +% by first converting the fractions cimals and then carrying oot the other arithmetic: 20333 %= 2.230 sum = 2.583 ‘Asa general rule the arithmetic should maintain atleast more digit in each fraction than is necessary in the Final esul. For example, if it is desired to describe the area of angle (area = width x length) to the nearest centime- er, then measurements of the width and length of the etangl should be made to the nearest tenth ofa centime- ef Use ofthe pocket calculator generally produces at least ation Numeric value 10°*, one quinlionth 10°, one quadrillionth 107, one titonth 10°, one billionth 10%, one milion 10°, one thousandth 10°, one hundredth 107, one venth 10',ten 10°, thousand 10%, milion 10° billion 10”, elton 10", quadion Mathematies and Statistics 3 ight digits of accuracy, which is more than enough far ‘nuclear medicine calculations, Percentages are values expressed as a fraction of some whole, entire value: 75% of some number is the same as (0.75 multiplied by that number. This is exemplified in the following: “T5%6 of Sm 75% Sml-= 3.751 Many pocket calculators have a % key, witich makes it unnecessary to frst convert the percentage to a decimal 25 enter percent value % percent key x utiply key 5 enter (ee result of 3.75 in display) Percentages are often used to express percentage change between two values. For example, a patient may have had a kidney fonction test last month that showed a kidney clearance rate of 42 milliliters per minute (ml/min). The patient returns today and has a kidney clearance rate of 58ml/min, The patient's kidney fonction has improved by 38% based on the following; 4% change = (new value ~ old valueVold value] x 100 68 ~ 42421 x 100 (6442) x 100 = 0.38 x 100 = 38% Algebraic Equations and Ratios Calculations in nuclear medicine often involve expressing ‘a mathematical concept as a ratio. For example, an intensity setting of 240 on a display screen may produce a suitable image for a bar phantom with 500,000 counts (e) in the acquisition Now it is desired to acquire an image to 300,000ct, The intensity on the sereen is known to change linearly with the intensity contro setting. Wat intensity setting should be used for cte new 300,000ct image? This translates mathematically into the following: x__ 20 intensiyy Books 500k This is typical of numeric equations in muclear medicine First, itis necessary (0 translate a mathematical concept, expressed in words, into an algebraic equation, The equation generally contains several numbers and ore unknown value; here the new intensity setting is the unknown value Go. The object is to solve for the unknown value by tear- ranging the terms in the equation. In this example the unknown intensity x can be isolated on one side of the equation by multiplying both sides of the equation by 300 ket. Remember, any math operation can always be done to both sides of an equation without changing the equality. Many students may have learned this technique as cross- multiplying to salve a proportion het x| —* vote 40 inten ] S00 Js4 Nuclear Medicine and PET/CT Technology and Techniques ‘The 300 ket cancels in the left numerator and denominator, 50, 144 imensity Another frequently encountered problem relates to radioactivity concentrations in patient doses. For example, ‘he morning elution of the "Mo.”*Te generator yields 943 millicuries (mC) of "Te radioactivity in 20ml of saline cluate, What volume should be withdrawn from the eloate vial into a patient syringe to perform a 20mCi patient scan {assume that no decay correction is needed)? x__ 20m 20mCi~ 40mC x=20mcix 208 50% — 25% + 12.5%, Attime = 0-9 ty Zao + Ha (See Figure 13 for a geaph of the radioactive decay law as 4 function of time for "Te with tia = 6 hours) Because both forms of the radioactive decay law are valid, it can be written as Aa" Aeon Because the expressions in the exponent ¢ in these equa tions are equal, it can be seen that a relationship between Aand ig is given by: Mathematics and Statistics 11 0.693 The A and t,p are inversely proportional to each other: a large A means a small (yp and vice versa. Whether to use tne form ofthe radioactive decay law or the other is simply a matter of convenience. Given 2, tia can be calculated easily (and vice versa). For example, given that the decay constant A for "Te is 0.1153h¢", what isthe tin? 0.693 __ 0.693 0093 — 6 ore 2 O1S3hr Careful attention to the units is necessary to avoid errors. ‘To make the units clearer, this might be restated as 0.693 T oss To get the units of I/Ar out of the denominator, multiply bboth mumerator and denominator by units of hours (essentially malyply by 1) 0.693_he Sasol oust The I/hr and the hr cancel in the denominator, leaving: 0.693 ‘e053 The common radioactivity prablem is to solve for one of the four variables (A, Ao 6 tin) when the problem specifies three of them. For example, on Monclay at 8 ax, a sample of "1 (Ga = 804 days) is calibrated for an activity of 10uCi, What amount of radioactivity will be given to the patient if the dose is administeted on the following Friday at 2 PM? Given: Ae = IOUCI, t= Mon B am Fri 2 oa, y= 804 days Solve for A= Acer1™* Here, a digression to discuss units is necessary. Revaiember the good practice of always writing down the units associated with every number Any problem involving the function must have a dimensionless number for the value ‘of x, so it is absolutely necessary that identical units be used for both t and ty. Then the units cancel in the numerator and denominator of tity, making the calculation independent of the units chosen to measwe time. In this problem the time of decay (Mon 8 4M —> Fri 2 PM) is 102 hours, but tua has already been given in days. It would be correct to express the time of decay ¢ as 4.25 days (rather than 102 hours) with the ty also in days, or it would be correct to express the ya as 193 hours (rather than 8.04 days) with the {also in hours, A= 1ONCi x 09 » (4.25 days/8.04 days)12 Nuclear Medion and PET/CT Technology and Techniques = Loycixeserannnnnn LoYCi x 8 = 693uCr A quick check of the results of the calculator’s answer is, also useful, based on the 100% —> 50% —+ 25%... time- line for each halflife. In this example the decay time of 4.25 days is less than one ty (8.04 days), so the answer should be between 100% and $0% of the initial 10-yCi actives The calculator result of 6.93yCi agrees with the mental cheek. Inadvertent calculator usage errors can be prevented with these mental checks A radionuclide is often calibrated for some activity level on a Friday, but it might have been administered to the patient om the previous Monday. This type of radioactivity problem can be solved using negative time values for times that precede the time of Ay calibration. For example, @ radionuclide with a 2-day halle is calibrated for Friday at noon to be 3mC3, What activity level was present on the preceding Monday at noon? Given: Ao = 3mCl Find Bm xe Ox ta 3imCi xe (notice [a] cimes [+] is 1) 2mCi This result is easy to check mentally since the time diffe ence is exactly 2 hall-lives; the answer should be that ‘Monday noon has four times the activity of Friday noon, in agreement with the calculator resul Sometimes a problem concerns only the fraction of remaining radioactivity (A/A), rather than the actual remaining activity in mCi. For example, what fraction of radioactivity is left ata time equal to 3 half-lives? Here, t= tua is given and the object isto find A/Ap. The radioactive decay law can be algebraically rearranged (dividing both sides of the decay equation by As) as: Alay = 0.125 A guick mental check confirms the result: in 3 hall-lives the radioactivity should decay 100% —> 50% > 25% — 12.5%. Remember that the radioactive decay equation always, refers to the remaining activity, which is the same as 100% minus the decayed activity. It is important to determine whether the problem is stating the ammount of radioactivity remaining, which the decay equation predicts, or the radioactivity that has decayed away. For example, the previous problem showed zat only 12.5% of the original radioactivity remains afier 3 half-lives. This problem could also state that 87.5% of the radioactivity decays away in 3 hatflives. Radioactive decay problems can also require the ¢ oF fz value to be solved for in the radioactive decay law. These values are contained in the decay law as part of the exponent function, $0 the exponent must be removed. An, cexamaple might be to calculate the time necessary for 99.9% of a sample of Te to decay away. Remember that the decay law works wich the remaining radioactivity 1% = 0.001 olhe Solve for typ tm AIAg= EOP" ‘0001 = e2@* oni ‘Taking the natural logarithm of both sides and using Inve’) =x yields: [n(0 002) = -0.693 x V6 01 br Now the ¢ value is out of the exponent and the equation can be simply rearranged algebraically 6.01 hr x In@.001)0.69% Notice how the minus sign and the units are carried cor rectly. Now using In(0.001) = 6.908 yields: 6.01 br x (6 908/0.695) = 59. 9hr The two minus signs are multiplied and cancel each other About 60 hours (or about 10 half-tives) is necessary for the radioactivity of "Te t0 decay to Yoo of its original value. For a radionuclide such as "", itis still true that 10 half- lives cause decay to about Yon ofthe original activity, but for "1, 10 halflives is 10 x 8.04 days = 804 days As another example, an experiment finds that a sample of radioactivity decays to 30% of its original value in 5 hhouts. What is the ti? Alby = e100 030 In(0 30) = 0.695 x (Sht/in) 1.204 = -0.693 x (Shit) Shr S208 = 0.693% =29hr Do the mental check. Does this seem a reasonable answer? 1f2.9 hours isthe correct fy, then a decay time of 5 hows is not quite 2 half-lives. So the expected answer is that 5 hhouts of decay with a 2.9-hour ga should leave slightly more than 25% of the radioactivity remaining, The problem thas sensible results: 30% remaining at time just less than. 2 hall-lives. In many practical problems ic is necessary to combine radioactive decay calculations with concentration-volume problems, The nuclear medicine department ray obtain its radioactivity from a Mo." generator through an eatly- morning elution of che generator. This eluate decays throughout the day, resulting in a changing concentration in mCi/ml), For example, a generator is eluted at 7 AM,900 mCi in 20m! of eluate solution. What volume be withdrawn from the eluate vial into & syringe to }I5mCi for a scan at 2 PM? First calculate the radioac- maining in the eluate vial at 2 Pt A= 9Q0mCi x ea x00 402mci mental check confirms the reasonableness of this slightly less than 50% remaining at atime slightly than ty) The concentration (radioactivity per fn the eluate vial is then ("°%yq) at 2 PM. The needed 10 be withdrawn into the syringe for a 15- at 2PM can be calculated from the equation activ- eoncentratioa X volume Aacxy Mind) ey 2a rsmei-( rsncie ey common problem for radioactive decay is to caleu- ratio of Mo (aa = 65,9 hours) activity 10 Te in generator chuate, The problem is that some "Mo iso eluted out along with the Te in the morning n The Mo isa radionuclidic impurity that is limited ory agencies to be less than 0.15uCi Mo per fe at the time of injection into @ patient. Consider erator that is eluted at 7 AM and yields an eluate vial ining 30uCi Mo along with 250mCi ”*"Te, Can this = 2HO gy HEM SomcP TE mc TE Omen TTR ATT = oncix<0.939 28.17"Mo nged very much in the 6 hours since generator elution, purthe "Te activity has halved, resulting ia large increase. inthe "Mo/""Te ratio, Mathematics and Statistics 13 Te "Mo activity rato ' 0.094 2 os, 3 025 4 032 5 039 6 oad 7 oa) 8 ost 10 bat 2 0.68 ra 073 16 078 18 80 2 083 2 oss 4 os? 26 oss 30 oat = 095 ‘The mathematics of this type of radionuclide generator are governed by the laws of radioactivity. relating radionu- cides denoted as a parent-daughter-granddaughter (and so con) decay chain. The parent radionuclide "Mo decays to the daughter **Te, which in turn decays to the granddaughter Te, and the decay chain continues. Without dealing with the exponential algebra for this type of decay, itis possible to easily calculate the "Te radioactivity expected to be luted from the generator by knowing three values 1. The activity of Mo in the generator, which is given by the manufacturer's calibration date and the decay law for *Mo ((yy = 65.9 hours) 2. The time since the last elution of the generator, which is commonly 24 hours for daily elutions 3. The ratio of *"Te1o "Mo in the generator, which depends fon the time since the last elution "Te to Mo ratios are shown in Table 1-3 as a function of the time since Ist elution) For example, a generator is delivered on Saturday and calibrated by the manufacturer for the following Monday at 6 Pat to contain 2Ci of "Mo, The generator is elured daily (Monday-Friday) at 7 AM. What activity of *7Tc is available in the generator at 7 a4 on Tuesday? The required data are as follows: 1, Mo activity Tuesday 7 aM, 13 hours after calibration An 2cixeammanasne ‘A= 1744mCi of ®Mo in the generator, Tuesday 7 4M14 Nuclear Medicine snd PET/CT Technology and Techniques 2, Time since I8st elution is 24 hours, si is eluted daily 3, From Table 1-3, the "*Te/?*Mo ratio is 087 for 24 hours since last elution, so: the generator Activity 87 x Activity "Mo 087 x 1744mCi Si8mct Depending on the quality ofthe generator, only a percent age of this 1518mCi of Fe will appear in the eluate. This is known as the elution efficiency of the generator. I the elution elficiency is 95%, then the Le Tuesday morning eluate would be calculated as 0.95 x 1518mCi = 1442 mCi, This is available for patient studies. Half-Life: Biological, Physical, and Effective fn most clinical applications the nuclear medicine gamma camera measures the radioactive counts over an organ of interest in the patient's body. Typically the patient's organ excretes the radiopharmaceutical with some biological half- life f while the radioactivity also decays physically with a physical halflife that is denoted as f. The biological half life isan indicator of the physiological fate of the radiophar- ‘maceutical, ty. The counts observed by the gamma camera, follow an exponential decay law based on the effective half-life t, where: or, ina format that is much easier for calculation purposes, Tett) ‘or example, i the liver exetetes a "Te radiopharmaceutical with ty=3 hours, then the gamma camera over the liver -would abserve an effective hal-life of = Shr 3H ake ‘ehe+ 30) Absorbing stsorton brn erivt Ve Soroaergetc beam Ineniy = be Thikrase Figure 12 The effective halb-life is always less than or equal to the smaller of fe 0F ty Attenuation of Radiation The calculation of the intensit (D of x-ray or y-ray photons ransmitted through some thickn rial follows exactly the same algebra as the equatic radioact igure 1-2 shows a beam of mon {getic stay or y-ray photons striking a thickness of absorbing material. Monoenergetic means that. the photons all have the same energy, such as a beam of photons from a fc radionuiclide source that emits photons with an energy of 140keV. The initial intensity (number of photons per second) entering the absorbing material is called 1h, material attenuates, orabsorbs, some fraction of the photons, and the photon beam emerges with a sransmitted (e., 1 absorbed) intensity I. The intensity of the transmitted radia tion is given by (0 of absorbing mate s for wis the linear attenuation coefficient, or the fraction of the beam absorbed in some (very small) thickness x. The linear attenuation coefficient w is the atalog of the constant Ain radioactive decay. ficient w depends on the type of absorbing material and the energy of the photons. A large 4 value means a strongly absorbing material. For “Te y-rays the 4 value in lea about 23cm", whereas the win water is only 0.151 Llead is a material that absorias much more strongly than water. A typical caleulation deals with the fraction i/lp trans- titted through a thickness x, For example, what percentage 140-KeV photons are transmitted through 10cm of water = 0.15em decay nuation coef The lines Because 22% are transmitted, it can also be sai are absorbed in 10cm of water, at 78% Leod climatic ‘Asenuted bam aod colimetar lersty = 1 ng mediuma problem thac asks what fraction of "I photons 364keV is transmitted through a half inch of lead 4¢ 364KeV), An initial inclination might be to 5 follows Ween iy # eter x00 calculation is incorrect because the units in the expo- dovnot cancel each other, A dimensionless number in the exponent to make the result independent of is used to describe w and x Any units can be used e410 look up the 4 values, from published tables, hen convert the thickness x to the corresponding father than convert the units of sto correspond with of For this problem, convert x=0.5 inch to 1.27em tr 1208 906) ‘of ead transmits only 6.196 of a beam of 364- ons from This also means that the lead absorbs fthe photons. The linear attenuation ceeficient the same problem as A; itis dificult to con- Its therefore common to follow the method radioactive decay and define a halt-value layer that thickness of material that absorbs 50% of the HVLis the analog of fn in radioactive decay. transmits 50% of the photons, two HVLs trans ‘of the original beam, and so on. The absorption like the following Layers petenronne lationship exists between 4s and HVL, given by L If the BVL or value is known, it is a rbrward calculation to find the other and uation equation is niost convenient, example, what percentage of 140-keV photons eare attenuated by 10cin of water? The HVL for tons in water is 4.6m, which- y= esonra Liga e800 Wly= 022, oF 22% yA fraction (0.22) of the photons is transmitted. s precisely the same result that was obtained previ- in this section using the attenuation equation with 4 problem, however, asks what fraction is aenuated by 78 is the attemvated fraction; Mathematics and Stavisties 13 10cm of water absorbs 78% of the photons emitied by "Te, This type of calculation can be carried out easily on the pocket scientific calculator. 0.693 enter value 4 change sign key x multiply Key 0 enter thickness value + divide key 46 enter HVL value = (ee resul of division) INVandInx ore key (see result of 0.22 in display) A quick mental check of the calculator result is useful, The thickness here (10cm) is slightly more than 2 HVLs (since HVL = 4.6cr), so the attenuation answer should be slightly less than 25%. Another problem might require solving the attenuation ‘equation for either x or the HVL, both of which are contained in the exponent in the attenuation equation, Just as in radioactive decay (where or ty needed to be solved for), the exponent with the natural logarithm can be eliminated as the inverse funetion of e+ p= er = e209 000 nou yx = 0.693 x (HVE) For example, if 10cm is known to transmit 22% of the photon beam, what Is the HVL? Jig0 22) = ~0.693 (WemV/HVL, ~0.693%10em tn@22) 0.693% 106m “151 Sen) Note how the minus signs cancel and how the units are carefully carried. To calculate the HVL on the pocket cal- cculator, perform the following steps: 0.693 enter value +h change sign key x rmltiply key. 10 enter x value + divide key 0.22 enter fraction value Inx natural logarithm key Gee result of 4.6cm in display) (One further nuance occurs for attenuation of photons. The value, oF the corresponding HVL value, for any material also depencis on the physical density p (glen; the HVL in ‘water vapor is different from the HVL in liquid water, witich isalso different from the HVL in ice, and so on, This makes calculations using the 1 or HVL values somewhat difficult because the 1 values are usually tabolated only for the common physical state of the material in question (¢g. for16 Nuclear Medicine and PET/CT Technology and Techniques liquid water). To circumvent this problem, a new parameter is defined as the mass attenuation coefficient fn (m"/g) = fp. The mass attenuation coefficient is independent of the physical density ofthe absorber and is therefore easily tabu. lated. The transmission of photons can then be expressed in one of three equivalent forms “ uM u Calculation ofa result using the more easily tabulated value cf mass attenuation coefficient also requires looking up the density (9) of the absorbing material It should also be noted that these equations for the transmitted fraction of photons apply only to a situation known as good geometry, or narrow-beam geometry, meaning, that a very thin, pencil-like beam of photons enters the absorber and is detected by a small collimated detector Scattered photons (photons not traveling in a straight line between the origin of the photon and the detector) are excluded by the good geometry. In most practical applications, such as photons arising in the patient's heart and being detected by a gamma camera, the geometry is broad- ‘bear. The photons can leave the patient's heart and move toward the thyroid, for example, and then scatter in the thyroiel and travel ward the gamma camera to be detected as a photon apparently arising from the thyroid. The scattered photons increase the apparent transmission through the patients body, and the equation is typically modified by a multiplicative buildup factor B, with B > 1. Buildup faciors are dependent on the absoréung material, energy of the photon, and geometry. A large patient has a buildup facior greater than that of «thin patient, so in broad-beam geometry Y= Bem Specification ofthe buildup factor, or Some other correction for scatter, such as arbtratily using a smaller value, may be necessary for accurate quantification of photons origi- nating inside a patient Graphs tm the modern technological world, people are inundated with data. Graphs provide a practical, highly visual way to convey large amounts of information. They can also be used to predict one vatiable based on another. The most common type of graph uses a linear set of x and y axes in the familiar Cartesian coordinate system. Figure 1-3 shows a graph or plot of the remaining radioactivity level versus time from a sample of radioactive material with ty2 = 6 hours. The x- axis (or abscissa) shows the time of each data point (or 0, 2, 6, and 12 hours), and the y-axis (or ordinate) shows the activity in millicurie for each data point. One variable is often considered to depend on another independent variable. The independent variable is generally plotted on the x-axis, with the depenclent variable on the y-axis. in Figure 1-3, time is considered the independent variable on the x- g 8 Actiy Porcentoge of raciooty 77 pee Time thous Figure 13 Linear plot of radioactive decay for ta = 6 hours, axis, and activity level is considered the dependent variable on the y-axis Mathematically, activity () is @ function of time, or y = {(0. The data points in Figure 1-3 are called discrete data points because a measurement of activity was taken at four discrete, individual data points (0, 2, 6, and 12 hours) Graphs of discrete data points may or may not, at the user's discretion, show the data points joined by smooth curves or a connect-the-dots type of straight line. Knowledge of the decay process suggests a smooth curve should best represent radioactive decay. On the other hand, a graph of the number of monthly kidney scans might be graphed as discrete data points joined in a connect-the-dots fashion because no reason implies a smoother curve Figure 1-3 shows a continuous curve of activity versus time forthe radioactive decay equation AzAe! The yo for the data in Figure 1-3 is 6 hours because the radioactivity drops 100%, 50%, and 25% in 0, 6, and 12, hours, respectively. If the natural logarithm is taken on both sides of the equation, the curve in Figure 13 will simplify imo the straight line shown im Figure 1-4 because In = laldae™) In = Indy + Inie) nA = Indy at or yrabbe This is the equation of a straight line with y-intercept a = Indo, and slope b=-A. This isan important practical result: the logarithm of radioactive decay data plotted versus time is a sttaightine graph with negative slope equal 10 the decay constant, Note that the logarctmic plot of activity still shows che activity dropping by 50% every 6 hours. ‘A brief review of the graphic interpretation of straight. line data seems pertinent. 4 straight-line graph of y versus x is represented by the general formula yeatbe8 8 88 xe3d loge of radioactivity 8 tT to Time thou) Figure 14 Semilog plo: of radioactivity for ty2 = 6 hours. The syaxis is logarithmic, Y=5+0.5(X) Ax=10 0 | 10 15 20 x Figure L5 Straightline graph with slope = 0.5 and intercepe = The y-intercept, which is the value of y at x = 0, is represented by a, The slope, which represents the steepness of the line, isrepresented by b Figure 1-5 shows a straight-line ‘graph with b= 0.5 as the slope. The slope is calculated from any two arbitrary points on the line as: 4y O2-w) Slope Ax G7) In Figure 1-5 the slope is calculated from the points Ge, 4) = ©, 5) and (Xo, 92) = (10, 20). 0-5) Soy 0-0) 0s Mathematics and Statistics 17 Figure 1-6 Two-cycle semilog graph paper. For a straightline graph, it does not matter which two points on the line are chosen to calculate the slope; the same answer is obtained. In Figure 1-5 the y-intercept a = 5 because the value of y is 5 at x= 0. Hence the equation of the straight line in Figure 1-5 is given by: y25+05x Given any x value, the equation can be used to calculate the y value. The sign of the slope merely reflects whether the curve slopes upward (a positive slope) or downward (a negative slope). Straightline curves ate generally used in ruclear medicine to prove a direct or linear relmionship between two variables or to predict some y variable based ‘on the value of the x variable, Mathematical relationships or curves that are nonlinear, such as radioactivity versus time, are often transformed by a mathematical operation to make the graph a straight line. For exponential curves such as radioactive decay, taking the natural logarithm of the activity transforms the data into a straight fine. The straight-line form might be preferred because subsequent interpretations or catculations ate simplified. The use of logarithms as a process to trans- form curvilinear data into straight lines is so common that a special type of graph paper is often used to simplify the process (Figure 1-6). Semilogarithmic graph paper has axes with divisions that are proportional to the logarithm of the ‘axis, so the y values are simply plotted at che appropriate point without the necessity of calculating the logarithm of the yraxis data. Semilogarithmic graph paper is available18 Nuclear Medicine and PET/CT Technology and Techniques with a varying number of cycles, or powers of 10, on the “y-axis. Figure 1-6 shows two-cycle graph paper, which can ‘accommodate y values that span a range of no more than 10%, The user simply relabels the numeric values on the {y-axis to correspond to the range of the data involved. For example, the y values could range from 0.23 10 79, with, the y-axis labeled with 0.1 at the bottom, 1.0 in the middle, and 10,0 at the top. Alternatively, the data might fall into the range of 200 10 8000, with the cycles labeled from 100 through 1000 to 10,000. Measurement of Effective Half-Li A typical procedure with nuclear medicine data is to caleulate the effective ty» of excretion from some organ in the body. As an example, consider a patient who is giver a meal of radioactive food to determine the tin of the emptying of the stomach. The radioactivity counts emanating from the stomach are plotied on semilogarithmic graph paper (Figure 1-7) The tyn of excretion can be determined by drawing a frechand visual-estimate straight line through the data points. Each data point is contaminated with statistical and systematic noise, or uncertainty, in the actual y value, 50 the straight line will probably not pass exactly through all, if any, ofthe data points. Use of this method to determine tna telies on the data poinis being reasonably well represented as a straight line on a semilogarithmic plot. Often the data appear more and more like a straight line at later time values, s0 these values are used to estimate the straight- line fit, Afier the straight line is drawn, the tye can be dletermined as the interval needed for the straight line to decrease by 4 factor of Ys Just pick any convenient starting value for p, then read from off the graph the time needed to reach fj In Figure 1-7 the visual-estimate best-fit straight 11 Log semach counts) — Bbal bof ne Time has a y-intercepe (or ys) of 7800cts, and the line falls to 3900cts (%4) in 12.5 minutes, $0 fia = 12.5 minutes by the visual-estimate best-fit line In some cases the data never decrease by a factor of 'h cover the time of the experiment, but itis still desired to calculate che ty. In this case the proper calculation (remem- bering radioactive decay follows the equation Iny = Inyo — 2) 5 to find the slope, which is equal to“, and the negative decay constant, which is based on the counts () and time (9 of any two points on the straight line Unty,)—Inly,)} be=t) Then the ty» is calculated from the typ = 0.693/a, Note that the equation for the decay constant slope requires calculation of the natural logarithm of the count values in che numerator. The logarithm need not be calculated to plot the data (because of the convenience of semilogarithmic graph paper), but calculation of the slope A does require caleulation of the logarithm of any two arbitrary points on the straight line. Suppose that the data were acquired through only 10 minutes and that it was desired to still caleulare th although the data do not decrease to (%) in only 10 ‘minutes, Using two points on the estimated best-fit straight line—(, 7800) and (10, +400)—the decay constant is calculated as follows: 0.573 Tome Tie (rites) Figure 1-7 Measurement of effective fy» = 12.5 minutes in the stomach by visual estimate of best-fit line on semilog plot= 0.693/4 is calculated as 12.1 minutes, in close with the graphic method, which estimated ty, = ‘mines, Note thet many nuclear medicine procedures yield gapas of coumts versus time that do not follow a simple ight line on semilogarithmic plots, Several half-lives or | organs might be excreting a radiopharmaceutical the body. Its then not correct to caleulatea single ta any data that do not appear to follow a straight line silt in different values for the fy2, which is based on the igh line, or in different values for predicting a y vale my x value using the straight-line fit to the data. A more matically precise method to fit the straight line to the isthe method of least squares, or linear regression. In technique a set of n data values at the points (x, 3. h > n, 18 graphed and fitted with a mathemati- llyexact technique, No mprecision occurs; every observer ‘of calculation is generally perlormed to show a linear jnship between two variables or to predict some y able based on measurement of some x variable. igure 1-8 shows data for x,y values from an experiment involving measurement of cardiac ejection fraction (EF) by Iwo different cechniques: a previously used method and the 43 +0.800% v0.94 4 Yn Low of identi 20 30 40 ‘Mathematics and Statistics 19 ‘new method, which involves some change in experimental techitique. Do these data show that the old method and the rnew method yield identical results? Or, given some value for EF by the old method (an x value), what would be the predicted results for ejection fraction by the new method (a ¥ value)? The least squares method, or linear regtession, calculates the best-fit values for y-intercept (@) and slope (b) in the bestfit straight line: y = @ + bx. The intercept and slope parameters define a straight line, as discussed previously. The least squares method calculates the a and b that minimize the sum of the squate of the distance in the y Girection between the bestfit straight line and the data points. Use of a scientific pocket calculator can greatly speed calculations. Typically, a calculator requires the user to enter the first x value and press some key (g. X; in Figure 11), telling the calculator this is xy; to enter the corresponding first y value and tell the calculator this is y, (with the ¥, key; See Figure 1-1); and to enter x), y2, and 50 ‘on, until all n data points have been entered. Then the cal- cculator displays the intercept and slope, Use of a computer ‘seven simpler, the user enters x, y values in a spreadsheet format and instructs the computer to plot the data and the regression line, Almost all popular spreadsheet and graph- ‘cs software packages offer linear regression (eg. personal computer [PC] spreadsheet program calculates and plots the best-fit line, Calculation of the regression line by hand is tedious but rot really complicated. The frst step is the calculation of four sums based on the x, y data values. Then some simple ‘multiplication and division produce the intercept and slope as follows. First calculate the four required sums, using the data for Figure 1-8, as shown in Table 14: 30 % 30 FF Old method 18 Regression analysis or least squares best-fit curve to compare old and new method for calculating ejection fraction (EB).20 Nuclear Medicine anf PETICT Technology and Techniques peri rene Old method: x (%) ‘New method: y (0) a 6 2 65 3° 4 3 32 8 56 2 6 3 56 6 38 28 uM 2 46 SHO OBIT OPK senda eroraf he estate BEB) = 59, Ex= x) best be T+ 6d4 +51 = 481 nyt tn SB EOS tee = aT tae hE Sa 48) + 62165) + aT 6245 $5146) = 24165 “Then the intercept and slope are calculated as ,-Exep-ex@y)] [nGx}-ExB0] 1245431479) ~ 4810516591 _ 45 o@4543)— 481081 and (Say) = Bxty01 EB] ‘The best-fit line is then given by = 945 40.800) This line i draven onto the graph of y versus x by caleulat: ing two points at the endpoints of che regression line and then joining these points with a straight line, At x= 20 the Destfit line value of y is 9.43 + 0.8020) = 25, and at x = 70 the best-fit line value of y is 9-43 + 0.8(70) = 65. The regression Tine has been drawn in Figure 18, Also shown in Figure 1-8 is the line of identity, which ts the line with a y-imercept of O and 2 slope equal to 1, defined by y = x. The line of identity is often drawn on regression graphs when the same parameter, here the ejection fraction, is being plotted on both re x- and y-axes. The line of identity facilitates an evalvation of whether the two methods (che x and y-axes) are predicting the same result for ejection fraction. I the two methods produce the same value for ejection Fraction, then the regression line should be the same as the line of identity, The line of identity is not 2 useful concept if the perticutar experiment being considered does not have identical parameters on the x- and ‘y-axes. A regression of blood pressure versus age, for example, Would not use the line of identity So what does the regression equation predict im Figure 1-8? The old and new methods do not predict exactly the same ejection fraction since the predicted regression line does not coincide exactly with the line of identity. For any given old method result (the x value), she regeession equation or graph can be used to predict what the new method would yield for an ejection fraction result. This predtetion of ay value, based on some measured x value, 1s the goal of segression analysis. Figure 1-8 shows that the new method produces results greater than the old method for ejection fractions below about 47 by the ok method because the regression line is above the line of identity. The new method underestimates the ejection fraction, compared ‘with the old method, for election Fractions greater than 47 bay che old method, Figure 1-8 also lists the value of goociness of ft param- ctets, which help detesmine whether itis reasonable that che data points are truly represented by a straight line, The lear correlation coefficient r is calculated from the sums used to find the intercept and slope, along with one addi- toonal stm from the data points (Ey nid) ExC8)) )=ExtEo}” [nlBy")- By yy] The sign of the correlation coefficient is metely the same as the sign of the slope. A correlation r= 1 15 perfect fit with the straight line passing exactly sbrongh each data point. The closer ris to 1, the more highly correlated are the.xand y data. A chance. or probability (P), always exists that daca that are truly not linearly relate! to each other will sandonely appear ina fatrly linear fashion. The statisti cal significance of correlation values less that | rmust be evaluated from statistical tables of probability (Table 1-5). If the correlation value for the n data points shows a P value fof less than 0.05 from the statistical tables, then the data ate sail to be linearly correlated, Figuve (-8 shows r= 0.94 for a= i data points, which from Table 1-5 has P< 0.001, so the data are highly linearly correlated, Ifthe r value was 0.632 or less the P= 0.05 value for n= {0 from Table 1-5), chen the data are not shown to be linearly correlated and it would not be prudent 10 apply a best-fit straight line t0 the data. A scatistically signifier: correlation (meaning an. r value with P< 0.05) merely implies a linear relationship between the x and y variables. This linear correlation or relationship does not necessarily suggest that x causes y. & graph, for instance, of the number of human babies (y) versus the number of stork nests (x) would probably show a high correlation, buc this does not mean storks bringFg eS re eee ee Tore foes rot PROBABILITY (P) D100 G50" OGIO G.005 0.001 0805 OBIS UNS GOTH OOO oss 0632 0765 O.8Ds ORT? Gast C514 Dem 0.583 0760 0578 044s 0561 0.802 0679 0306 031 0463 0499 0.570 Bes. More likely, both x and y depend on some other able, such asthe number af people in the house: peaple We babies, peop live in houses. storks baile) nests on te houses, “The use of Microsolt Encel for linear regression is quite worward. I 10 paits of x, y data are Known, as in =SLOPEWBLBIO.ALATO van cell C2 and type in SINTERCEPT(BL-BIO,ALAIO) Eick on cell C3 and type in SCORRELOLBIDALAT that Excel then calculates slope. intercept, and correla- oefiientinio cells C1, C2, and C3, The results ean te priced, ie x,y data can be changed, and so on. IF dat values charge, note that the values for slope and ‘tems will also change. The user can also quickly ask celto plot the data: highlight the block of cells containing x,y data by painting a box around them by holding gn the let mouse button, Then chick on the Chart Wiansd gmear the iop of the desktop and select XY-Seatter for the of plot, Aller the plot appears, ask Excel to add the regression line 10 the graph by doing te following: in tie center ofthe graph, click the word Chart atte line wil be added io the graph. eciher goodness of fi parameter shown in Figute 1-8 the standard error of the estimate (SEF), sometimes ted S,, whict is defined as follows: SPE = JEKy a — bY n= 20” SEE \s the root mean squate average deviation in the y ction ofa data point from the regression line—that is, fo far away the average data pot is from the regression en the » direction, An Figure 1-8, SEE = 3.9 so the ge data point is about + units on the y-axis from the cession line. Another imerpretatinn of SEE might be what Mathematics and Statistics 22 the old and new ejection fraction methods elisagree by 4, con average, A line that passes exactly through each data point produces r= 1 and SEE =0 for a perfect fit. The SEE rows larger as the data points fall farther from the regres sion line, There iso easy way ta state a P value for whether the SEE is small enough or not small enough 10 say the ft 4s scatiscioaly acceptable Least squares uve filting could also be used to improve the precision of measurement of effective iy of data such as that shown in Figure 1-7. The calculation meee requires using the time as x values and the In (counts) as the y values in the regression equations, The best-fit line hasa slope that is the decay constant 4, and the hall-life is dhe tus = @.659/A, This least squares fit eliminates the imprecision of visual-estimate best lines Several other caveats are known regarling least squares carve fitting. The linear correlation coefficient r deseribes only the deprce of linear relationship betaveen x and y. Some ‘exact nonfiear Felationship may i Tact exist between x and. tha yields a nonsignificant linear correlation. Data gener ated from the relationship y = 7 (= -2, ~1, 0, 1, 2 with cottesponding y=+, 1,0, 1, 4) would produce the straight- line fit y =0.0 + 2x, wh r= 0. An exact relationship in fact exists beaween x and y, but it is not linear The least squares bestfi Line intercept and slope values hnave 4 standard error uncerrainties, according to formulas available in stmistics texte." This results in a range of possible best-fit lines. A statistical uncertainty is actually in the ‘predicted y value forany x value. These uncertainties in the predicted line can be used, forexample, toascertain whether the ptedicted regression live is really different from the line of identity, orthe and y values ean be tested for difference ‘using tests, as discussed in the following Section. Other Graphs Another graph useful in muclear medicine is & histogram, which is used to display a single list of numbers such as the number of tests per month (Figure 1-9). Note that the vyaxis Is displayed with a mininnunn value of 2000 rather ‘than 0 to aceentuate month-to-month differences. Another type of graph is the pie chart, which is typically used to present some whole object broken down into its component parts. The width of each slice of the pie is proportional to each component's percentage share STATISTICS a Mean and Standard Deviation ‘Statistical analysis n nuclear medicine allows lor consicles= ation of such things as the level of confidence in the aecu- rey of some measurement or the level of confidence in a patien’s test value differing from a normal result. Two concepts that often emerge t statistical considerations are precision and accuracy. Precision refers to the spread, or range, of data values obtained when some parameter is rmeasuied many times oF i many patients, A small preci-22 Nuclear Medicine and PET/CT Technology and Techniques 2000 2750) Nan to 3 2250 2000 Dy FMAMI GAS ON DI Month Figure 1-9 Histograt showing monthly variation in nuber of tess, sion means that hitle variation exists im the data values Accaracy,on the other hand, refers to how close the results are tothe true, or “gold standard” result, Suppose that the length of a 96-inch-long board is measured twice, obiaining, results of 94 inches and 98 inches. These data cauld be called accurate because the average (or mean) value of the two yields the correct answer of 96 inches. The large range of the data (04 to 98 inches), however, would be considered rnot nearly precise enough for any practical construction measurements. Another person might measure the length ‘of the board and obtain 96.25 and 96.50 inches. These results are more precise (smaller range of data) but less accurate (the average 1s 96.37 inches) than the previous measurement, & good measurement is both accurate and precise, but no necessarily fixed relationship exis between accuracy and precision in any experiment, Statistical parameters, such as the mean and standard deviation, allow quantification of the concepts of accuracy and precision Suppose that some parameter x is measured n times. Then the mean (or average) of then values ofthe parameters xis defined by the symbol: The symbot ¥x means the sum of all the measurements of x Asan example, a patient has four measurements of serum thyroxine (Ty, thyroid hormone) given by 9.5, 9.0,9.2, and 8.7upél, The mean value x is calculated as follows: 05490492487) 4 ug/l Ifthe true answer for the serum T, was known, then the ‘mean value X= 9.1 ughll could be compared with the true value 0 discuss the accuracy of the test. The standard deviation, « measure of the precision of the data, is given by the symbol 6, defined as = (Be ~ Hin DPF Recall that an exponent of ' means che same as the square root (W"), The standard deviation is a measure ofthe deviation or spread of the data points from the mean value. For the thyroid T, data 0= (95 ~9)* + 0-91" +2-91" +8 7-9'Y a-u = (00.16 + 001 +001 + 0.16151" 6 = (03473) = 01133! = 03366 ‘9 = 03 gil, coreectly rounded to one significant decisal place ‘The standard deviation is often expressed in a format of the mean value +o, so this patient has a T, test result of 9.1 O3pigidl. A large @ value indicates data with a large range ancl therefore poor precision. Sometimes it is more useful to express standard deviation asa percentage of the mean value, which is frequently called the percent standard deviation, or coefficient of varia~ tion (CY): For the thyroid T, data above, 0.3367) ny Sig )*100= evel ‘The CV merely shows how large the standard deviation is when compared with the mean value, For example, consider two separate experimental measurements ofthe length of an object. Both experiments produce the same standard deviation of 2cen, but the two objects have different mean lengths of 2m and 2km. So one experiment has a CV of (@em/200.em) x 100, or 1%, whereas the other experitnent hhas a CY of em/200,000cm) x 100, oF 0.001%. The standard deviations of the two experiments are the same, but the CVs are vastly different, showing a much better precision in the second experiment (CV = 0.001%) than in the first experiment (EV = 1%) Microsoft Excel can be used to calculate statistical parameters. As an example, for the four values of Ts, enter these data values in cells Al, A2, A3, and A¢ in a new Excel spreadsheet. Then, in cell Bl, ype =AVERAGE(AT:AS) In cell B2, type: DEMALAS) In cell BB, type 7B Note the coefficient of variation is calculated in cell B3 as (0.036995, and the user must realize that this is 3,6995%, properly rounded to 3.7%,cther example, consider two different blueberry Suppose Farni A has blueberries of diameter 6 id Farm B has blueberries of diameter 9 + 21mm he scientific pocket calculator generally has Keys for ation of mean and standard deviation. The user enters fis data value and then pushes the key eg. X) that the calculator this sx. Then ehe second data value is ed and the X, key is pushed again, telling the calculator 3,4 soon up t0.X,, Then press the key to display sean % and o Gometimes denoted on calculators as nthe calculations with statistics require consideration e theoretical distribution of the data values. The user may w that a sample of blueberries produces diameters of 9 ‘but suppose the user wants to know how many erties have diameters greater than L1 mm or what che bility is of getting a blueberry bigger than 13mm. The tical distribution refers to a theoretical measurement il the blueberries of Farm A ot of the Ty values in all nt, and so on. The distribution is a theoretical graph low many blueberries there are with which diameters or ‘any patients there are with a certain value of T, is usually measured in an experiment is not the theo- Alstibution but rather a sample of data values (say 20 erries or 10 patents) that is part of the whole theoreti- distribution. Two types of theoretical distributions of ¢¥alves are important in nuclear medicine: the Gaussian sian distribution if a reasonably large sample is taken avery large distribution, where the parameter being ured is expected to vary randomly, For example, the that something follows 2 Gaussian distribution ns that if the number of times a certain value occurs is Mathemavics and Statistics 23 Number oftets “a 2 6 «F Fe 420 430 2 8 = Mom F 8 § i : Lok $f ¢ Test results Figure 1-10 Gaussian distribution for test results can be ‘expected 10 show the results in large population where test value is expected co vary randomly. Note that 68% ofall cest results are within the mean #1 standard deviation Cz * 16), 95% of all test results are in the range of § + 26, and so on plotted on the y-axis versus the measuted numeric value (eg. blueberry diameter or number of counts) om the x axis, then a graph such as Figure 1-10 is obtained. Analysis reveals that a Gaussian distribution tas 68% of all results within the range ¥ £ Lo, 95% of all results within ¥ + 20, and 99% within ¥ # 30, Figure 1-10 shows this distribu tion, Hence returning to the sample of blueberries, if 5mm and ¢=2 mm, then ifall the blueberries in the bucket are examined, the following would be expected: 68% within 7 to Limm (& + 16) 95% within $10 13mm (% +20) 99% within 3t0 15mm (Z+3) ‘The chance of getting a blueberry less than 5mm, or greater than 13mm, is only 59. This is called dhe area in the tas (the ends) of the diseribution. Because 95% are within ¥ 2g, it must be that only 5% are beyond the range ¥ + 20. ‘The properties of the Gaussian distribution are often ‘used in quality control procedures and to determine normal ranges A quality control procedure might measure some daily parameter to determine whether an instrument is working properly. If today’s value is beyond the ¥ + 2 value (based on previous measurements), then the instra- ment might be considered to be operating improper!y because the probability of getting a result beyond & + 205 only 5%, This type of quality control, or normal value range determination, uses the 95% confidence interval (the ¥ 2a range) as a conventionally accepted range. This means that a} in 20 (or 5%) chance exists that an acceptable quality control test will be incorrectly rejected, or a 5%24 Nuclear Medicine and PET/CT Technology and Techniques chance exists that a normal subject would be incorrectly called abnormal. The preponderance of results, 19 out of 20 (055), will he correctly classified Poisson Distributions and Counting Statistics Counting statistics, meaning the number of counts expected from a sample, fellow the Poisson distribution, In the Poisson distribution the standard deviation (0) for any number of counts (C) is fixed at the square root of C: This fixed definition of standard deviation does not exist im Gaussian distributions; essentially only counting stat nies ae Poisson, One farm could have a Gaussian distribution of blueberries of diameter 9 + 2mm (ie, mean + 0), and another farm could have blueberries of diameter 9 44mm, Neither farm has 6= V5. Any combination of a mean value and a standard deviation is possible in Gaussian data, ‘whereas Poisson counting statistics always have the standard deviation of the counts from a sample equal to the square root of the counts EXAMPLE: What isthe standard deviation ofa sample that produced 10,000 counts? 9c= VIGO: EXAMPLE; Woat isthe coefficient of variation of 10,000 100 counts 100 10,000 100% =15 a EXAMPLE: What are the 0 and CV for 100 counts? 9, = ¥100 = L0counts 100 a-( 20038 = 10% “These examples show that as the number of countsincreases from 100 to 10,000, the sbsolwe magnitude of che standard deviation goes up (from 10 to 100 counts), but the percent standard deviation or CV goes down (rom 10% to 1%). Less relative variability occurs in large count values. Counting statistics, besides being Poisson, are also described by a Gaussian distribution as long as the number of counts is greater than about 30. Hence given some number of counts C, the standard deviation is automatically known, [t is known that 68% of repeat measures of the sample fll within C+ YC and 95% of repeat measures of the sample fall within C # 2V, and 59 on, EXAMPLE: What is the 959% confidence interval for & sample with 10,000 counts? 95% confidence = C £ 2a = C + 2VE 10,000 # 210,500 = 10,000 + 200 counts The 95% confidence interval is then from 9800 to 10,200 counts. If this represents a sample of the counts actually ‘emanating from radioactive material, can be said they are {95% sure thatthe true counts from this radioactive materlal are in the 9800 to 10,200 range (itis expected that 95% of repeat measurements would be in the range 9800 to 10,200 counts). The number of counts actually emitted by a xadio- active source varies from time to time, in a random fashion, according to both Poisson and Gaussian distributions. The variation in the mumber of counts is often relerved to as counting noise. Sometimes this type of problem is expressed by saying that 10,000 counts are needed to be 95% confident tha the true count is within 2% of the measured value, The ptes- ence of two different percentage values can seem confusing The 95% confidence interval on 10,000 counts is +200 counts, and this figure of 200 counts represents 2% of 10,000 counts. The general formula expressing the number of counts needed wo be no sure that the true answer is within some percent () of the measured counts is given by: (bs) In this formmala, n is replaced with a 1 for 68% confidence, 8 2 for 95% confidence, and a 3 for 99% confidence. EXAMPLE: How many counts are needed 1o be 95% sure that the true answer is within 1% of the measured valle? Here n=2 and p= % Pe {2 100% IC this sample were recounted, then 95% of repeat measure- ‘ments would be in the range 40,000 + 2.J40,000 or 40,000 4400, which represents a 19% uncertainty (because 400 is 1% of 40,000), Sore radiation counters may have a preset count dial labeled as the percent error at the 95% conf dence ievel, sather than a preset count dial actually labeled im number of counts, Setting such a preset count dial at 1% error would be the same as collecting 40,000 counts. Setting such a preset count dial a 5% would mean collecting 1600 counts, since Some problerns may deal with the count rate R that is obrained by counting a number of counts C in timeThis presumes that no error occurs in measurement of the time Z By using C= RT, the og can be alternatively written as tbe following 6 EXAMPLE: A sample shows 43,627 counts in 5 minutes, ‘What ae the count rate and standard deviation in the count rte? eis) T 44627 counts min 8725 counts _ R =8725epm o 2epm 5 Hence the count rate would be expressed as 8725 Aepm, EXAMPLE: The count rate is 876Scpm for a 2-minute count, What range of count tates would be predicted to occur in 68% of repeat measurements? This is simply aang what is the value of 0 because 10, is the 68% «confidence imerval, 9,= MINT) = VTE = VIBES = 66 epmn Ifthe experiment were then repeated, it would be predicted that 68% of the repeat measurements would be in the range 8765 + 66epm Background courts are a problem in most measurements ‘of counts from a radioactive sample. Background arises from natural terrestrial and cosmic sources of radioactivity ‘ot from other nearby sources of radioactive material. The sample is uszally counted in the presence of some background radiation, yielding a gross count for sample plus background, denoted by the letter C. Then the sample is removed from the counter and the background B is counted, The net, ot true, counts, which represent the sample only, is denoted by N and given by: N=C-B ‘The standard deviation of the net counts is given by: y= VCD “This is an example ofthe rules for combinations of errors\ ‘phic state that for addition or subtraction, errors add in quadrature within the square root ose) Aativionlly, since o: = VE and oy = VB, one obtains n= Nic=B) EXAMPLE: A sample is counted in the presence of back round yielding C= 1600 counts, Then the background is counted alone, producing 900 counts, What are the true or net counts and standard deviation? Mathematics and Stauisties 25 N= C= B= 1600 ~ 900 = 700 counts 6, = V(1600+ 900) = VI5DO = 50 counts lence the net counts can be expressed a5 700 + 50 counts. Note that the presence of background has increased the statistical uncertainty in the results. If chere were no background, the sample in this example would have produced. N= 700 + V700 = 700 + 26 counts Sienilar considerations can yield the rules for the net count rate Ry = Re~ Ry=C/Te = BIT oy = Me RFE R) om = MCTTEF BIT) or om= MR TEFRTIS Chi-Square Tests ‘Suppose the wish isto test the reliability of the operation fof some counting instrument, a5 a quality conerot test, by confirming that the instrument consistently produces the same counts in a reliable fashion. How can this be done twhen itis now known that the number of counts recorded will vary from measurement to measurement because of the statistical nature of radioactive decay? The answer is to count a sample a reasonable nurmber of times (typically 10 repeat measurements are used) and then determine if the 10 different values show an acceptable or reasonable amount of variation, Too little oF too much variation in the repeat measurements indicates a counter that may not be func- tioning propetly. The chi-square test (7') is used to derer- mine an acceptable range_of vatiability in the repeat measurements, The mean of the 10 measutements is Gexermined (C = ZC), and is caleulated as: Table 1-6 shows the calculation of z* for some counter, which yielded a value of 247 for x’. Is this an acceprable amount of variation Between the 10 measurements? The definition of aoveptable 1” is given by looking up the prob- abilicy @) of this 7" value for the n (here 10) repeated ‘measures in Table 1-7. Too little variation is shown by a 2? smaller than the P = 0.9 value; an amount of variation exactly as expected from the statistical nature of the decay process would yield P= 0.50, and an unacceptably large 2° ‘would yield a P-value greater than 0.10: 1f 0.90 > (P of x") > 0.10, detector is OK. If @ of 7) > 090, the detector shows too litle variation, If @ of x) < 010, the detector exhibits 100 much variation For the data in Table 1-6, 2? was calculated = 24.7. Ia Table 1-7, this n= 10 measurement has a P value < 0.01, s0 this detector is not operating properly, if the zis too large, the26 Nuclear Medicine and PET/CT Technology and Techniques detector is showing too much variation ia counts, The detector probably needs to be repaired. An acceptable 2° for 10 measurements would be in the range 417 P > 0.10 range. It is only common convention that fixes the P = 0.90 — 0110 range as acceptable, Some laboratories right be willing to accept more variation, or they may be willing to use 0.95 — 0.05 as the acceptable P range forz Notice the similarities in statistical approach for ealeu- lating the linear correlation coefficiert ¥ and? Both are calculated trom the data based on some equation, and then the P value of the r or 72 is determined to assess the si nificance of the result. This is typical of most statistical tests calculate a parameter and then look up the probability, of this result Does the result for the data in Table 1-6 seem sensible, that the detector is showing too much variation? Consider that a properly operating detector should have 68% (about Square Observation Counts (C) c-08 1 lone 447 299381 x 10285 1057 mans 3 sar 3323 1104233 4 woes 13, 17 5 W382 a7 298253 6 24 AT 2,902.1 7 9986-1933 37,3649 8 10139 403 1.6241 8 10356 W767 31,2229 1 im 3 39153 101,793 E=0 B= 251,666 Tatas 1672 7 out of 10) of repeat measurements within the mean count + sfimeancount). So the deviation C~ 7 in Table 1-6 shold be less than 1a, less than + VI0I793=101 counts, in 7 out of 10 measurements. Examination of Table 1-6 shows that only four of the measurements (instead of the expected seven) have deviations from ihe mean within le, or within 101 counts. Too many measurement values ate beyond ‘tla. The detector is not working properly by this erade visual check, in agreement with the excessively latge value. Similar visual checks can be made to assess that only cone aut of 20 measurements should have deviations from the mean of more than #2 x 101 counts. tn fact, Table 1-6 shows that one measure is a deviation of ~333 counts about 30) and another is almost 2e (actually -193 counts), so bout 2 out of 10 are beyond 42e in contrast with the predicted 1 out of 20 measurements with this mek variation ‘The calculated P value of 77 will always agree with such commonsense data checks Lastly, word about calculating 74; most scientific pocket calculators do not have a calculation funcuon for 72, but ‘most caleulators can calculate the standard deviation of the repeai count measurements, denoted here as SD: EC. " PSS] This can be algebraically solved for (C= €: y= (=) x GDF ‘This is simply the numerator of the definition of 77, so the equation for 2 can be written in terms of the mean and SD as the following wor p=) To calculate 7, therefore, he pocket calculator (or the PC computet spreadsheet program) is used as described previously to calculate the SD and the mean, which are then used in the above equation for 7’. The data in Table 1-6 show SD= 167.2 with a mean @ = 10179.3, so (107.2 101793 ae road exactly as obtained previously in Table 1-6. ENIESSEIS pe Number of sample measurements 99 095) 090" 5 030 on 106 10 209 333 “7 1s 4.99 ost 1 20 763 101 uu? "Salle acepuable 7 Ish this column "anges acceptable 7s i this column, PROBABILITY (P) 0.50 210) 005 oor 336 178 949 B38 at yr 169 22 BB m1 237 291 183 22 301 362To use Microsoft Excel to caleulate 77, using the data of “Table 1-6, just enter the data into cells A through A1O in anew Excel spreadsheet. In cell Bl, type the number 10, or the mutmber of data points. In cell Cl, type = (BI-D*STDEV(AL:ALO)“2/AVERAGEAL 10) Excel then places the value of 7? in cell C2 The 7 statistic, in other algebraic definitions, has other uses besides atest for stability of counters in nuclear medi- «ine, Chi-squared is commonly used in 2 X 2 contingency tables which could consider, for example, the number of petsons who are well or ill, tabulated according to being male or female. Alternatively, a categorization could be mace of the number of persons with positive or negative imaging tests using gamma cameras from two different ‘manufacturers. A 7° calculation would show whether the two gamma cameras find different or same numbers of positvernegative test results, Tests Thettestisused totes for differences between mean values.” This testis also commonly referred to as the Student t-test because the original proposal for this test was published by «an author using the pseudonym Student, The two forms of the tes are for independent samples and paired samples, For independent t-tests, two independent groups of data are used. For example, some test result is measured in a group of ill patients and a group of nz well patients, The two groups are completely indepenclent. No relationship or concation exists between the two groups, typically because the gtoups represent different patients. The two groups have mean and standard deviations denoted by %, SD,, and SD, The¢-estisa test ofa null hypothesis thatthe mean values inthe two groups are equal. The hypothesis being tested can be abbreviated Hy: %, = %, The null hypothesis is statistical nomenclature, meaning that no difference is found. Thistest is called a two-tailed test because either group could be ‘equally expected to have the langer mean value, The caleulation of the test requires no new algebraic details from the data, the t-test result is simply calculated from the means, ad standard deviationsin the two groupsas follows: [o—vsso2+00,-nxs02}(22) tq-ayf kaa) ‘The calculated 1 is then compared with tabulated values of the critical rtest (Table 1-8) with ny + ns ~ 2 degrees of freedom () at the P = 0.05 level 2S tyesus then the study does not reject Ho: & no statistically significant difference is shown between the two groups, IEt> tao then the hypothesis Ho: 5, = % is rejected, and the difference between the two groups is said to be statistically significant. Mathematics and Statistics 27 EXAMPLE: One group of ny = 6 patients is given a drug, and their thyroid T, levels (68, 8,7, 9.2, 8.6, 8.5, 9.0) have %, = 8.80 and SD, =0.26. Asecond group of m= 5 patients 's given a placebo, and their T. results (8.3, 85, 82. 81, 4) have ¥, = 8.3 and SD; = 0.16. ls there a difference in T, levels between the two groups? Did the drug affect T, level? (8.80-8.30)/ (6-n%026'+6-««0167)+(43)]" le-wxea6'se-nncorey}e(5+5) oD 030 oa. 03e 3 a er 01794)" From Table 1-8 the critical ¢ value at the 5% or 0.05 level with m, +m —2=9 degrees of freedom is tooos = 2.26. Because calculated tis greater than critical, it can be concluded that @ statistically significant difference exists between the drug and placebo groups. The drug did make a statistically significant difference. In fact, the statistical tables show the critical t statistic atthe 19 level tan01= 3.25, thus demonstrating more than 99% confidence that a difference exists between the drug group and the placebo group because calculated tis greater than tygoy. Itcan also he said that less chan 2 1% likelihood exists that the difler- cence between the drug and placebo groups is the result of a chance occurrence rather than because of the drug Most PC-type computer software packages (spread- sheets, statistics software) offer calculation capabilities of Ee z= y? Degrees of | PROBABILITY (P) freedom (9) 0.10——«0.05* 010.001 4 213 278 490 861 5 201 237 403 687 6 Loe as an 596 7 199236 350 Sal 8 1.86 231 336 504 9 183226 3250478 0 Lal 223 Bay 439 re 175 213 295 407 20 in 2.09 28+ 3.85 30 170 204 275365 60 167 2.00 266 346 = 164 196 258 329 “Minimum vale oF forsgnifican diflerence sin his colamm Larger values imply a more sgnfican ciference28 Nuclear Medicine and PET/CT Technology and Techniques t-tests for which the user simply enters two columns of data values and selects an independent or paired test. The PC calculates the test result and the associated probability, sparing the user from any tedious calculations. In using such software it can be helpful to enter some test data, such as from the examples given here, to ensure that the software produces the same ftest and probability values as in the examples given here. Alternatively, the t-test can be per formed as in che example here, alter first using the pocket calculator to calculate the means and standard deviations of the two groups as discussed previously. To use Microsoft Excel for an independent t-test, for the Ts data example, ‘enter the first group's set of data values in column A (cells AL, AZ, AS). enter the number of data points for the first group in cell BI (i.e, for the T, example data, enter the number 6 in. cell BD, enter the data values for the second {group in cells C1 through C5, and enter the number 5 in cell DI (representing the number of data values in the second group). Then click on cell El. and type in STE? BDI) Click on cell E2 and type in: TTEST(AY:AG.CLC52,2) Im the Excel equation above for TTEST, the numbers 2, 2 are ielling Excel to do an independent two-tailed (est Excel then places the t-statistic (573) in cell El, and the probability P (0.00466) for this (statistic in cell £2. 11 P< 0.05, then Hi is rejected and a statistically significant difference becomes evident between the two groups. If P > 0.05, then Ho cannot be rejected; a statistically significanc difference between the two groups was not found. In this example, the P value of 0.00466 is <0.01, so Ho can be rejected even more stringently at the 1% level, while being, more than 99% sure that the differences between the two ‘groups are not caused by chance, just as concluded by the hand caleulations for the P-value being < 0.01 special term, standard error (SE), is used when referring to the variability in meam values. The SE governs the variability of repeated measurements of the mean value, wheseas, the standatd deviation (SD) governs the variability of any cone individual patient measutement. The result ofan experiment to measure a mean value usually is expressed as § SE, rather than % +t SD, although this is not universal practice and conlusion ocewss i the specification of SE or SD is ‘rot made clear. The SE s given by. SE=SD/Va_ The dats in the Ltest example above would most cotnmonly be specified as mean + $ Drug group mean = 8,80 £ 0.11 yp Placebo group mean = 8.30 + 0.07 yg/ll Stating the data in this manner helps compare the means inthe two groups. A graphic representation often shows the individual data values as dots on the graph, with a separate symbol for the mean. The mean symbol on the graph often includes + SE bars, drawn as vertical lines of length equal to 1 SE above and below the mean value. The testis essen: tally a measure of how far apart the mean values are measured in terms of their standard errors. A paired Hest is used for testing dllerences between two ‘mean values when the dara are from the same patient. For example, a group of n patients could have their ventricular ejection fraction measured before (xi) and alter (x) taking some drug. The ejection fractions before and alter drug ‘administztion ave correlated; a patient with 2 large ejection. fraction before drug administration could also be expected to have a large ejection fraction after drug administration What matters in paired data is the difference d = x, ~ x between any two values. The difference between the first and second measurement is tabulated in each patient to calculate the mean difference ja Beurm) The standard deviation of the difference (SDD) is defined in the following manner Efe] sep | Note that the algebraic sign, either + or - of the difference (@ in each patient must be carefully used in calculating d and SDD. Then the paired (test tests the mull hypothesis that no mean difference exists between the first and second, measurements, Ha: 4 = 0, by comparing the caloulated stitistic, €= d / (SDD/VA), to the critical { value at the 0.05 level with n= 1 degrees of freedom (= tesoos) It > tives. no difference is shawn between the Wo measurements. If 1 bo1005, statistically significant difference exists berween the first and second measurements, Ifthe calculated ¢is just greater than the critical ¢ with m ~ 1 degrees of freedom at the P confidence level (€g., P= 0.05, 0.01, 0.005, 0.001, and so on), then it can be said that a diference is shown between the two measurements at the P confidence level Table 1-9 shows calculation of d and SDD for the old versus new ejection fraction data from Table 1-4, Note carefully how the algebraic sign (tor ~) of dand d ~ dis consid. ‘ered in Table 1-9, The mull hypothesis being tested is Hy: ole EF = new FF, The calculated t= 0.14 is much less than the critical value (90s 2.26) from Table 1-8, 50 the conclusion is that no statistically significant difference exists between the old and new EF values; Hy cannot be rejected, This is another way of saying that the tines regression line in Figure 1-Bisnotstatistically different from the line of identity, To use Microsoft Excel fora paited test, proceed to enter the data values exactly as described above for using Excel for an independent t-test. For the example data of Table 1-8, there would be 10 data values in column A and 10 data values in column C, Cells BL angé D1 would both be entered 5 10, but now enter the Excel forimulasas follows, Click on cell EL and enter: =TINV(E2,B1-1)SEIT S00=[0-F/we—0"] =[r96/00-0") =11996)"=447 salso0(s) 03/47/19) Chick on cell £2 and enter: =TTEST(AVAI0.CL:C10,2,1) ‘The 2,1 in the Excel TTEST equation here is teling Excel todo a two-tailed, paired t-test. Excel places the t-statistic (1) in cell E1 and the P value (0.89) in cell E2. Becanse Pis >0.05, the null hypothesis cannot be rejected; no dif ence berween the two éjection [raction methods was 1, which is exactly as concluded by the hand calcula tion method previously. Note that a t-test (or either independent or paired data) ther [ils to show a statistically significant difference (¢ 2 tertcal) ot does show a statistically significant difference it < tentical) A test that fails to show a difference does tut prove that the two mean values are equal,” it merely Indicates that the data are inadequate to prove a difference exits A semantic nuance that may be encountered is that pendent tests consider the difference between the eans, whereas a paired ¢-test considers the mean of the alrences more than two mean values are involved, the simple (Hest is not appropriate. For example, oné might wish to [wow whether a difference exists between the means of test value n four independent groups of patients, The null yparhesis 18 Hs. K, = ¥,= ¥, = FO. This test requires 4 hod called analysis of variance (ANOVA)—the most mrerful way to find differences between the four groups INOVA also olfers various methods for testing nat. just for hether all the means are the same bat also for differences: tween pairs of the mean value. Alternatively to ANOVA, less powerful for finding differences, the data can be ted for all pairs of differences with a t-test using the ferromi method, which cequites lowering the significant value from 0.05 to 0.05/n, where m is the number of pos- ‘Mathematics and Statics 20 EF new method Difference (d= old ~ new) and ava a 8 4 38 Lesa 2 6s 3 32 1024 » 4 ~ 52 2r0t 3 2 1 08 ost DS 36 3 -12 ist 2 6 3 32 1028 oo 38 3 28 7a4 6 58 68 46.28 28 at ~6 3844 aL 46 48 sible tests, For four groups of patients thete are six possible [Huests: group 1 versus 2, 3, nd 4; group 2 versus 3 and 4 group 3 versus 4 All six ttests can be done and with significant difference claimed only between pairs of mean values with a ¢ value larger than the critica t vale st the (0.0576 = 0.0083 P value Medical Decision Making ‘Medical decision making would not be necessary f medical tests always produced a result that correctly identified the patient as either ill or well. Unfortunately, because of the diversity of biological response, possible statistical noise 18 data, and other technological deficiencies, many medical resis can often produce identical results for ill and well patients. Therefore by 3 somewhat arbitrary method, a normal test resull range is established for well people and then results outside this normal Tenge ace said. to indicate illness. For example, suppose Ty thyroid hormone is measuted in a group of well people, and itis found that the mean £2 SD normal range is from 4 to 12ngjal. Any subsequent patient test would be defined as nortnal ot euthyroid for 42 42 12 and hyperthyroid for T, 2 12. However, isnt using an absolute upper limit of 12 to define an abnormally high T, value a bit dogmatic? A patient with T, = 11.9 is labeled well, whereas a patie with Ty = 12.1 6 labeled ill, even though the difference in the cwo values may be within the limit of accuracy of the experimental cechnique. This is because most test values indicate a contintium af results, Hot ayes or no answer, as in pregnancy, Waat about a patient ‘mage test that produces an image that looks suspicious, that {s,nor quite clearly normal but nat extremely abnormal? The image test can be calle either positive (yes, it shows defect) cot negative (no defect is shown), What are the consequences30 Nuclear Medicine and PET/CT Technology and Techniques fbeinga lax image reader (Ge. one who calls lots of positive ‘Studies) versus a strict image reader (Le., one who requires very strong abnormal image characteristics before calling an image positive)? These are some ofthe questionsthat medical decision making can consider. Atest that finds a result for which the test was performed is called a positive test. A positive test for hyperthyroidism would be a T, > L2ughil. A positive myocardial perfusion imaging test would be a patient image with a cold defect in the heart, A positive *°Te bone scan would be a patient image with a hot spot in the bone. The test result might be low, high, cold, or hot, but itis positive for the condition, for which itis being tested ‘A test thet does not show the tested-for result is called a negative test AT, result in the euthyroid (normal) range would be a negative test for hyperthyroidism. A myocardial perfusion scan with no defects in the heart image would be a negative scan for coronary artery disease. Note carefully that the postive or negative value refers to the test result; postive or negative is not defined by the patient's clinical condition, The patient's cordition can be similarly categorized as either well rill. This categorization must be obtained from some test other than the one being considered as the nuclear medicine test with positive or negative results. The patient condition is usually established as well or ill through some other medical test called the gold standard test. A patient for example, could have either a positive or a negative myocardial perfusion imaging test, and the true state of the patient as being well oil, is usually established using some other clinical test suck as coronary angiography. An imaging phantom study could have an image result that is, called either postive or negative, and the true gold standard result is the known fact about whether the phantom contains an imaging abnormality or defect. Some presentations of this subject material might use the terms normal and abnormal in lew of well and ill Given these definitions, all of the following material relates to placing each patient’ test result into one of four categories 1. True postive (TP): persons with a positive test result who are truly ill 2. False postive (P): persons with a postive test result who are not ill 3. True negative (ND: persons with a negative test result who are truly well 4, False negative (FN): persons with a negative test result who are not well A woman with a positive pregnancy test who is actually pregnant is called a TP result Ifthe pregnancy test result ‘was negative but the woman is actually pregnant, then the result is an FN result, More typically, in nuclear medicine a decision is made to call an image test positive if he image shows some agreed-on defect. Then patients are categorized as TP, FP, TN, or FN by combining their nuclear medicine image rest result with that of some other gold standard clinical test, A patient with a negative myocardial perfusion {maging test (no abnormal deiects on the image study) who is known to have coronary artery disease would be called an FN result. A phantom study that calls the image positive in an area of the phantom where a lesion definitely exists would be a TP result Now consider the following test results. A study of imaging tests in 1000 patients produced the following data Test Result ) w Well 836 = TN 44 FP ul 26 = FN 94=TP Several other parameters that define the usefulness of the test can be derived from these categorized data ‘The sensitivity, or true positive fraction (TPE), is the peroentage or fraction of ill patients who have a positive test: te Sensitivity = TPF. cP) For the above data, Sensitivity = oF =78% Ow meaning that the study found a positive test result in 78% of the ill people. Note the distinction between similarly appeating symbols, TP and TPF. ‘The specificity, or true negative fraction (TNF), is the percentage or fraction of well patients who have a negative Specificity For the above data, 236. Specify = =95% Bor meaning that the study found a negative test result in 95% ‘of the well people. Again, note the distinction between similarly appearing symbols, TN and TNE. (The decision ‘on whether to use the term sensitivity or TPF and specificity ‘or TNF is purely arbitrary) The above data show a sensitivity of 78%, which is ess than the specificity of 95%, meaning that this test does a better job at correctly diagnosing well people than it does at correctly diagnosing ill people. The ideal medical test would have sensitivity and speci- ficty both equal 100%. Unfortunately, medical data usually appear as in Figure I-LI, with an overlap between the test results for the well and ill patient populations. Which is ‘more important, high sensitivity (.e., correctly finding ill people) or high specificity tie., correctly finding well people)? The answer to this question is complex and often depends on the medical, economic, and social consequences of the test, IF the imaging test is intended to detect brain ‘tumors that are known always to be fatal, but for which a simple medical cure with no side effects can be found even in well people, then a high sensitivity might be preferred,Number of persons 0 50 are ether TN or FP, and ill persons who are either TP or FN. The consequences of misdiagnosing an ill person would be death, whereas the consequences of misdiagnosing a well person would be a harmless drug regimen. On the other hand, suppose the consequence of finding a positive result in an imaging scan was immediate total frontal lobotomy, ‘which would prolong the patient life by only a few months. Given the dire consequences of treatment, a high specificity ‘would be preferred to avoid misdiagnosing well patient. ‘Another factor that affects the most desirable sensitivity and specificity of a test is the prevalence of disease, The prevalence isthe fraction or percentage of il persons in the study population (IP+FN) Prevalence = ——CIP#FN) (TP=FN+TN=FP) The preceding test result data had a prevalence of (25498) 2849) yng, Qero4+ 839044) meaning that 12% of the study subjects were ill The prevalence is characteristic of the patient population; prevalence is not a characteristic of the test itself Prevalence is shown graphically in Figure 1-11 by the area ‘under the ill curve compared with the area under both the “land well curves, The relative heights ofthe ill and welll carves show the prevalence, which can vary from one hospital to another. Ifthe prevalence is high, many sick people “are present, andl the optimum test may be one that gives priory to a high sensitivity at the expense of lowered “specificity (Le, a few FP errors are more easily afforded in the small nurnber of well people). To the contrary, a screen- “ing test for some rare disease operates in a patient population with alow prevalence, wire the optimum test may be Criterion level Test (+) <| > Test (+) Mathematics and Statistics 31 Tes! result value Figure 111 Distrtbution of test resus for vo groups: well (D~) and il (D4). Tes sul is called negative (T=) or posive (7+) depending on whether wi below or above rterion Level. Note thatthe well nd ill populations overap, creating well persons Who ‘one that gives priority to high specificity at the expense of lowered sensitwvity to avoid a large number of FP results| in the preponderance of well people 2 parameter that specifies the total number of correct answers, regardless of being ill or wel. is the accuracy of the test {otal number correct) ‘Tora number of peesons) cIN+ 70) TN IP+EP+ FN) ‘The test result data show an accuracy of: Accurac (836494) _ gs, (i494 4442) meaning thatthe res correctly diagnosed 93% of ll patients as ill of well. One problem is that the accuracy of a test depends strongly on the prevalence of disease for the same sensitivity and specificity of a test. This means that the accuracy between two hospitals can be very different for the same test because the prevalence at the two hospitals can be very different. Accuracy can he writen specifically in terms of sensitivity, specificity, and prevalence as: Accuracy = (Sensitivity x Prevalence) + Specificity * (l~ Prevalence) where prevalence is expressed as a fraction and not a percent Using these parameters for defining the goodness of a test, consider the following situation. A patient has a posi tive test result on some nuclear medicine study, Hence this ‘most be either a TP result in a truly ill person or an FP result ina truly well person. The relerring physician simply ‘wants to know the probability of disease, given the positive32 Nuclear Medicine and PETICT Technology and Techniques test result. The referring physician can be provided a wealth, of parametets about the goodness of the test: sensitivity = 78%, specificity = 95%, prevalence = 12%, accuracy = 93% However, these parameters do not answer the clinical question at hand: what is che chance that this patient, who had a positive test result, is truly ill? A parameter is needed that deals with the clinical usefulness of the test, not just goodness-of-the-test parameters such as sensitivity “The postive predictive value or predictive value of a positive test P(D+: T+) can be read as the conditional probability of having disease (D+), giver a positive test result (T#). This is merely the fraction of persons with positive test results sho are craly il IP c+) P(DH:T 4) For the test result data, oF POHTH om ‘meaning that the answer to the referring physician's question is that only a 68% chance exists that the patient with the positive test result is truly ill. Of course this means that, 2 32% chance exists that the patieat with this positive ‘muclear medicine test result és nt ill. These considerations are a reflection of a mathematical concept called Bayes theorem, after Thomas Bayes (1702-1761), an English math- ematician and theologian. It allows recalculation of the probability of illness based on the new information of @ positive test result, Before the nuclear medicine test, the chance that the patient had disease was simply the prevalence (often referred to as the pretest probability of disease) of 12%. After a positive test, however, the probability of disease has jumped to a posttest probability of 68%. Conversely, the negative predictive value or predictive value of a negative test (DT) is the probability P of not having disease (D-), given a negative test result (T=), ar the fraction of all persons with a negative tesr who are truly well: WN TN+EN) For the test result data, P(D-:T-) = 83610836 + 26) = 97%, meaning that a patient with a negative nuclear medicine test result has @ 97% chance of being uly wel. Note Bayes theorem at work here. Before the nuclear medicine test, the patient had a (L-prevalence), or 88% chance of being well (ie, the pretest probability of wellness was 88%) Alter 2 negative nuclear medicine test, the posttest probability that the patient is truly well jumped 10 97%. The discussion so far has not touched on how che patients true condition, wel or ill, is determined frorn some gold standard test, Calculation of all che parameters (such as TN, FP, sensitivity, specificity, or accuracy) tequites that it is known from some other test, who is truly ill or well Hence seten evaluating the goodness of test results such as, sensitivity, itis always necessary to ask what was used for the gold standard to define illness. Two hospirals may have vastly different. parameters for sensitivity and specificity, not because of any difference in the nuclear medicine test result, but rather because the two hospitals define illness in differen terms. For example, an invasive procedure such, 4s coronary angiography is often used as a gold standard to calculate the sensitivity and other parameters of nuclear medicine myocardial perfusion imaging, but there may be rnp commonly agreed-on definition of illness im the gold standard test. One hospital might say that the test must show coronary arteries narrowed by only 40% for a diagnosis of illness, but a second hospital might say that a patient is not ill until the gold standard coronary angiography west shows more than 70% narrowing of the coronary arteries, This difference in the gold standard combines with, the characteristics of the nuclear medicine test to produce different sensitivities and other parameters at the two hospitals When calculating sensitivity, or example, itis necessary to carefully define the test criterion level that characterizes, 4 positive test. One hospital may say a positive test for hyperthyroidism is any T, > 10.Sug/ll, wheteas another hospital may say Ty > 12. Syg/dl is a positive test result Different sensitivity values result. Infact, when information from the diagnostic test is received, any criterion level can bbe chosen to define a positive test. A lax criterion level (eg, calling a nuclear medicine study positive with just a hint cof abnormality in the image) leads to high sensitivity and low specificity: Ifa high sensitivity ts desired, all tests can be called positive, resulting in a sensitivity of 100%. Of course, there would then be no TN tests, so the specificity ‘would be zero, Ifa high specificity is desited, all tests can be called negative, resuting in a specificity of 100% (but unfortunately, the sensitivity would be zero) As the criterion level that defines a negative or positive test result is moved lefeor right in Figure I-11, the sensit ity and specificity change accordingly. Sensitivity and spec- ificty then are not parameters fixed as constants in nature by some inherent property of atest or by an image reader's skill level, Rather, sensitivity and specificity depend on the gold standard used to define illness and on the criterion level that defines a positive test. In Figure 1-11, ifthe criterion level i slid to the right so that itis stricter in calling a positive test result, the sensitivity decreases while speci- fieity increases, Slide the criterion level to the left so that it becomes more lax in defining a positive test, and the sensitivity improves while specificity worsens, For further guidance when thinking about these matters, a receiver operating characteristic (ROC) curve can be plotted, which shows sensitivity and specificity as a function of the criterion level used to calla test postive ‘An ROC curve is shown in Figure 1-12. The y-axis is the sensitivity, and the x-axis is the specificity. Note that the canis is inverted in the sense that it decreases left to right from 100% to 0% specificity. (An alternate labeling for the axes of an ROC curve will have true positive fraction (TPE]sare 25 | tte oh a Specilicity % Figure 112 Receiver operating characteristic (ROC) curve showing sensitivity and specificity asa function of the imimutn image scone needed to call an image postive. Nove that he xaxis fr specificity reads left-to-right, from 100 to 0 lax reiler might call any image with a seore 22 positive, with high sensuvty and low specifety. A strict reader would quire a high score ro call the image positive, resulting m low itty and high specificity, The best overall test operates ‘ear the upper let corner of the ROC curve, ftom 0 to 1 on the y-axis, and false positive fraction [FPF] fom 0 10 1 on the a-anis) The ideal point for a test result ‘would be the upper eft comer of the ROC ettrve in Figure 1412, with sensitivity = 100% and specificity = 100%, ‘n ROC curve is a graphic representation of the eflect ‘ol changing the criterion level, just as the criterion level ‘vas moved let or right in Figuse 1-11. A wealth of information can be gleaned from an ROC carve concerning which sensitvity/specifcity isthe optimum operating point, that 1s, the optomurn etiterion level that should be used to call atest positive. No definitive answers may be available, but the ROC curve clearly shows the available options, The ROC curve is still dependent on prevalence and the gold sendard texto define illness. Suppose that aplanarimaging, study is reported in the literature with specificity = 61% and sensitivity = 86%, Another study in che literature reports a tomogtaphic study with specificity = 82% and sensltivity = 64% Which is the better test: planar or tomographic? The question of which testis better cannot be answered Luness both sensitivity and specificity are improved in one athe tests in fact, these data could represent the same test results; the difference could simply be from a changect criterion level for calling a test result positive. The hospital with te planar data could be calling a test result positive with @ lower degree of image abnotmality, leading ta Jmproved sensitivity but worsened specifiy. t cent be clearly determined which testis betier without seeing the entice ROC curve for both the planar and tomographic data Mathematies and Statistics 33 To form an ROC eueve ftom arty data, simply tabulate the TP, EN, TN, and FP results to calculate sensitivity and specificity for each criterion level that defines a positive test For an in vitro T, blood test, just calculate sensitivity and specificity wing diferent criterion levels for a positive test Gg, calculate sensitivity and specificity for T defined as T20, as T2 1, as Ty 22, and 50 on). Then a plot of ser sitivity. versus specificity ‘at the various criterion levels yiekds the BOC curve, Image data are somewhat different because a quantifiable, numeric result for imaging studies is not always available. Hence a rating scale 1s typically employed to score ant image frort 1 to 5 as follows: enage Score Meaning Definitely normal image Probably normal image Possibly abnormal image Probably abnormal image 5 Definitely abnormal image Suppose an imaging study with 50 patients is conducted. From some other gold standard test, itis known which, patients ate truly ill or well: 28 are well, and 22 are ill. A reader scores each image as above. The sensitivity and specificity can be calculated from the TN, FP, FN, arid TP results for each criterion level as follows: Positive TN FP FN TP. Specificity Sensitivity test 8, e) seore > (x) 0 28 0 2 oO 100 2 9 1 1 2 32 86 3 yon 3 6 86 4 23 5 8 oF 82 64 SGD 27 1 16 6 96 a7 Figure 1-12 is an ROC curve for these data. The planar and tomographic results shove could represent just two different operating points on this single ROC curve. The planar test could have beet called abnormal for an image score 23, whereas the tomographic data may have been called positive tes only fora stricter image score 24. Alter natively, a difference in ROC curves cou be noted for the planar and tomographic results, but this could be ascer- tained only from seeing the entire ROC curve for both studies. ‘A medical test cannot be judged solely on sensitivity and specificity because these parameters are a strong function of the definition of the test criterion level used 10 calla test postive, The ideal operating point on the ROC curve could be at the upper left corner, with sensitivity equaling 300% and specificity equaling 100%. In general, the test that is closest to this ideal isthe beter test, so the best operating ‘point (ae what score to calla test positive) may be the one that results in an operating point closest to the upper Tel. corner of the ROC curve. Alternatively, as discussed prev ‘ously, considerations of prevalence of disease anc consequences of diagnosis may suggest favoring sensitivity over specificity, or vice versa. Its also cotrmon to se twa di ferent ROC curves characterized by the area under the ROC curve, denoted by A,. The highest A. value is used to deter34 Nuclear Medicine and PET/CT Technology and Techniques rine which test is best, because a higher A, value means that the ROC curve is closer to the ideal operating point in the upper left corner of the ROC curve. ETN © ‘Scientific notation allows very lange and very small ‘numbers to be more conveniently represented. © Learning to perforin sequential calculations with a pocket calculator can save significant time in performing calculations in the clinic or laboratory. ‘© Many applications in nuclear medicine technology use ratio relationships. Dilution problems are examples of the application of ratio relationships: CV, = Ca¥ ‘The inverse square law proportionality k tek ¥ results in the intensity of radiation exposure f decreasing to one fourth when the distance is doubled. Units of radioactivity are the becquerel (Bq = 1 dps) or the curie (3.7 x 10° dps) © Conversion between taCi and MBq is accomplished by multiplying by 37 MBa/mCi. * Exponents are described in the general form base" umber. ‘The general form of the radioactive decay equation is = Age, or it may be expressed as A= Aye?™>*0) The decay constant £ and tare inversely proportional to each other: 0.693 ta R Decay factors (DF) are calculated from the exponential postion of the general decay equation DF = &8>*0# and are used to compute tables of DF values. The ellective half-lie ty 1s calculated from the biological half-life ¢y and the physical half-life t by the equation ‘© The intensity [of a radiation heam of high-energy photons is attenuated by the relationship 1 = le A hall-value layer (HVL) 1s the thickness of a particular material to absorb one half ofthe intensity of a beam of radiation, Fach additional half-value layer recluces the beam again by one half Stanclatd deviation is a measure ofthe precision of data and is given by the symbol o, defined as o= face - 3A - DIY ‘The randomness of ratioactive decay follows Poisson statistics, and the standard deviation is calculated by ve Tine coefficient of variation (CV) is defined as The standard deviation for counting applications in the presence of background radiation is calculated by MoD ‘The chi-square test evaluates the reliability of consistent operation of comnting equipment by evaluating variability in measurements Sensitivity is the true positive fraction, the fraction or percentage of ill patients who have a positive test Specificity is the true negative fraction, the fraction or percentage of well people who have a negative test Accuracy is the fraction or percentage of patients being correcily diagnosed as ill or well PESTON 1. The radiation intensity from a point source of Te is Omar at 3m from the source. Ifthe distance is changed to 9m, what is the new radiation intensity? a. 0.5mR/hr b. Lomm/br ©. 20 mR/he 4. 3.0mR/he 2. A technologist is sitting near a bone mineral densitometer, which isa point source of x-rays. If the x-tay beam intensity at Lm from the x-ray bbeam is 0.20mR/ht, what distance from the x-ray beam should the technologist move to decrease total weekly exposute to the occupational limit of 2m for 2 40-our workweek? a. 05m b. Lom © 20m 25m 3. 10mCi is equal to how many becquerels? a. 370Bq b. 370kBg . 370 MB, 4d. 370GBq, 4. 20mC1 is equal to how many becquerels? a. 0.74GBq b. 037GBq . 054GBq . 0.20GBq5. The dose equivalent for occupational, whole-body exposure is commonly limited 10 50mSv. How many rem is this? a, 5000rem b. 500rem: ©. 50rem, d. Stem 6. A source of " ftya = 8.05 days) is delivered to the ‘nuclear medicine department calibrated for 100mCi at 8 aM on Monday. I this radioactivity is Injected rmio a patient at noon on the following, Tuesday, what radioactivity will che patient receive? a. 80mCi b. 85mCi ©. 90mci a. 95mCi 7. A patient was injected with "4 on Monday at 10 aM. On Tuesday at 10 4M the thyroid probe was placed over the thyroid and 100,000 counts \wete produced. On Thursday at 10 aM the probe showed 25,000 counts, What is the effective hall- if in this patient's chy? a. 12 hous b. 24 hours ©. 48 hous a. 72 hous 8. A*Mo/Te generator is eluted Monday at 7 aM, producing 1.8) of "Te in the eluate vial, in 20m saline. What volusne of eluate should be withdrawn ftom the eluate vial into @ syringe an otder to inject a patient with 20mCi of "Te at 3 FN (given the (12 of Te is 6,02 hours)? a. 056m b. 066ml ©. O7om! 4. 0.86mi 9, If the HVL for some rastionuclide in Tead is 0.30mm, what thickness of lead shielding is necessary t0 reduce the radiation exposure fromm Bmi/he to TmR/h? a. 030mm b. 045mm © 060mm 4d, 0.90mm Mathematics andl Statistics 35 10, The linear attenuation eoefficiem in lead for “Te ‘gatnana rays (140 keV) is 23cm" What percentage of these gamma tays will be absorbed by a lead apron that contains 0.60 im of lead? a. 75% b.50% ©. 25% 412.5% 11. A new gamma camera/computer system that uses a new method of calculating cardiac ejection fraction (EF) is imstalled in @ nuclear medicine deparument, The department decides to caleulate EF for the next 25 patients on both the old ‘gamma camera and the ew gamma camera before discontinuing the use of the old camera. In the future, if tis desired to convert the new EF value to that which would have been obtained on the old gamma camera (eg. to assess ifthe patient's EF had changed), the mathematical analysis to be used is called a. Independent t-test D. Linear regression ©. Standard error 4. Chi-square 12. What is the standard deviation of 40,000 counts? a. 4000 counts b, 2000 counts ¢. 400 counts, d. 200 counts 13. What is the coefficient of variation of 40,000 counts? a. 2% b. 1% 6 05% 4. 0.25% 14. How many counts should be acquired into each pixel of a muclear medicine flood image if we wish to he 95% confident that the true count in each pixel is within 1% of the measured counts in each, pixel? 100,000 counts 40,000 counts 10,000 counts +4000 counts aoe36 Nuclear Medicine andl PETCT Technology and Techniques 15. A patient’ thyroid is counted with the chyroid probe and produces 8000 counts. Then the patient is removed and the background is found to be 2000 counts, The (net counts) + (standard deviation it the net comnts) in this patient is 4, 10,000 + 100 counts b. 10,000 + 77 counts ©. 6000 #77 counts 4. 6000 + 100 counts 16. The gamma camera seems to be producing erzatic results. A Co flood source is counted 26 times, producing the following count values (1000, 975, 1032, 1096, 982, 997, 1012, 1090, 994, 977) What is the 7° value for these counts? a. 193 b. 183 © 173 163 17, Which expression desctibes the operation of the ‘gamma camera in question 16? Operating properly b. Showing too much variation . Showing too little variation d. Not enough information to answer the question 18, One group of 20 patients is given a drug that is thought to have an effect on kidney function, and another group of 20 patients is given a placebo Ge., sugar pill, which is known not to have an effect on kidney function). The nuclear medicine jgamma camera is used to calculate the glomerular filtration rate (GFR) in these two groups of patients. What would be the proper statistical test to use to test the hypothesis that no difference in GFR exists between these two groups of patients? a. Chi-square test b, Paired ttest €. Independent t-test d. Linear regression analysis 19. A nuclear medicine test produces a positive test result in only 80 of the 100 patients known to be ill Similarly, the test produces a negative tex result in only 190 of the 200 patients known to be not ill, Which is correct? a. Sensitivity 95%, 80%, specificity accuracy = 90%, prevalence b. Sensitivity = 90%, specificity accuracy = 90%, prevalence €. Sensitivity = 90%, specificity accutacy = 90%, prevalence = 33% 4. Sensitivity = 90%, specificity = 80%, accuracy = 70%, prevalence = 33% 20, 2 2. 23. m4, One radiologist is known to be a lax reader compared with another radiologist who is known to be a strict reader. How would one expect their sensitivity and specificity 10 compare? a. The lax reader would have lower sensitivity and higher spectfciy. b. The lax reader would have higher sensitivity and lower specificity. ‘e, The readers would be expected to have the same sensitivity and specificity. 4. There is not enough information to answer the question A®Mof?*Te generator is calibrated for 1.00Ci of Mo at 7 aM Monday. The generator is eluted daily Monday through Friday at 8 aM, bur che ‘workload is especially heavy on Friday so the generator is eluted again at 3 PM to obtain more Te, Assuming 100% elution efficiency, how many mCi of "Fe will be eluted at 3 PM Friday? a. 930mC1 b. 70mCi €. 160mCi a. 80mci A radioactive source of "Cs (ta = 30 years) was calibrated on 10/23/2000 to contain 1042Ci. This 1s used as a daily accuracy check source in the dose calibrator. Presuming the dose calibrator is working properly, what activity should the dose calibrator show on 4/23/2006? a. 96uCi b. 88yCi © 54nCi a 22uci A source of °F (ia approximately 2 hours) is noted to contain 3raCi at noon, What was the radioactivity at 8 am that same day? a. 24mCi b. 18mci © L2m¢i ad. omc The biological half-life of "1 in some particular patients 30 days, The physical halflife is 193 hours. lf the patient’ thyroid is counted with the thyroid probe detector, what effective hall-tie will be observed? 26 days 6.3 days 5.1 days 12 days25, A adloactive source decays from 20mCi to 2.5mCt in 18 hours, What is the physical hall-life? a. 8 hows: b. 7 hours ©. 6 hours a. 5 hours 26. A satnple shows a count rate of 36,000¢epm during a J+minute counting period. Express this as the count rate + standard deviation, a 36,000 £ 220¢pm b. 36.000 # L1depm ©. 12,000 £ 55epm a. 12,000 £ 28epm 27. Alongetived radioactive source is counted for 1 imine and yields 10,000 counts. If this source is counted immediately again, there is a 95% probabilisy that dhe result wll be in the range: a, 9950 to 10,050 counts 1, 9900 to 10,100 counts 6. 9800 to 10,200 counts 4. 9700 to 10,300 counts 8. A S-mnl sample ofa standard diluted 110,000 ‘produces 27,200 counts in the well counter. A San sample of patient plasma, courted for the same time as the diftted standard sample, produces 99,100 counts. What is the plasma yolume? . 108L 274 1BL | OTL, 29. Readers ate encouraged (o use Mictosolt Excel wo solve this problem. Wat is the mean of the following five numbers: 6.40, 720, 3.50, 9.20, 510? QoL 628 597 1 483 6. Readers are encouraged 10 use Microsoft Excel to solve this problem, What is the standard deviation ofthe data in question 29? a. 402 b. 3.26 © 215 d. 1.92 Mathemauies and Statistics 37 31. Readers are encouraged to use Microsolt Excel to solve this problem, What is uxt coetlicient of variation CV of the data in question 29? a 39% 31% 28% 1% aoe 32, Readers are encouraged to use Microsoft Excel to solve this problem. In a nuclear medicine technology training program, the stuclents wonder their final exam grade in their training program is related to their subsequent boatd score on the NMTCB exam, The five students atained the following pairs of scores (exam score, boatd score) = (72, 60, (84, 74), (8. 71), 68, 60), (1, 79), Wha is the linear correlation coefficient r between cexatn score and board score? a 099 b. 095 ©. 086 ao7 33. Using Table L-5, what is che P value for the linear correlation coefficient rin question 31? a. P= 0.001 b, 001 < P< 005 © 003 010 34. Using Table 1.5, does a linear correlation coefficient r= 0.95 for n = 5 daca poinis indicate a statistically sigosficant linear relationship between the x and y variables? a. Yes. b. No ¢. Not enough information to answer the question 35. Readets ace encouraged to use Microsoft Excel to solve this probleme. What is the slope and intercept in the regression equation for problem 31? (board score) = intercept + (Slope) x (exam score) a. intereepi = 7.15, slope = 0.502 i 15, slope = 0,602 15, slope = 0.702 15, slope = 0.802 a. imiereepe 36. Using the regression equation in question 31. if the training program exam score was 84, what does the regression equation predict for the board score? 92 8 72 a pose38 Nuclear Medicine and PET/CT Techoology’ and Techniques 37. A thyroid uptake dose of LOC is equal to how ouch radioactivity m St units? a. 37kBq b, 370KBq 37 MB, 4. 37MBq, 38. What is the proper number of significant figures for the quotient (6.1 x 10™)/(0.1232)? a. 049513 b. 04951 «0495 4.0.50 39, What is the proper number of decimal places for the sum (0.3) + 6.5264? a. 3826+ b, 3826 © 383 438 40, Ie is known that 75% of a radionuclide decays away in 12 hours. If we start with 10mCi, how much will be left after 6 hours? a. T5mCi b. 5mci ©. 2.5m @.1.25mci Ea 1. Bevington PR: Data reduction and error analysis for the physical sciences, ed 3, New York, 2003, McGraw-Hill 2. Bushberg JT et al: The essential physics of medical imaging, ed 2, Philadelphia, 2002, Lippincote Williams and Wilkins. 3. Colton T: Statistics in medicine, Philadelphia, 1975, Lippincott Williams and Wilkins.CUB oe Tac TeCy CHAPTER OUTLINE SAECTROMAGNETIC RADIATION ATOMS AND MOLECULES Fectrons protons Neutrons ‘ATOMIC STRUCTURE MASS-ENERGY RELATIONSHIP Mass Detect Nuclear stability NUCLEAR NOTATION AND NUCLEAR FAMILIES ECAY PROCESSES ‘Alpha Decay Beta Decay ¥ TERMS Describe the structure of Positron (s") Decay and Election Capture Gamma Decay ‘SCHEMATICS OF RADIOACTIVE DECAY Mathematics of Decay RADIOACTIVITY UNITS Decay Calculations erage Hal-tife and Effective Half-Life Parent-Daughter Radionuclide Relationships INTERACTIONS Interactions of Charged Particles with Matter effective halflife cleetromagnetic radiation electron eapeare gamma decay healftife hhalf-value layer (AVL) sober ssomer OBJECTIVES After covmpleting vhis chapter, the reader will be able to + Describe the properties of electromagnetic radiation + Desc atom and its components swat Excitation and Ionization Alpha Particles Beta Particles Annitilation PHOTONS Photoelectric Effect Compton Scattering Pair Production ‘EXTRANUCLEAR ENERGY RELEASE Bremsstrahiung Radiation (Characteristic X-Rays ‘Auger Electrons [ATTENUATION AND TRANSMISSION OF PHOTONS isotope linear attenuation coefficient swuclear stability pair production photoelectric effect positron decay radioactive decay be the imeraetions af charged particles with and their properties, Explain the struccare ofthe eh define the line of stability: Sate the relationship of mass and energy in Einstein's equation Write the correct form of radionuclide notation. * List the nuclear families and state theit characuristcs Name and describe the primary forms of radioactive decay. Diagram the schematics of the various radioactive decay process Define decay constant Use the general form of the radioactive decay equation to calculate precalibration and postcalibration uantites of radioactivity ist the radioactive units and becquerel Write che equations for average halt-life and effective halflife and calculate elfective andl biological of the nuclides and + Discuss the processes of excitation and ionization ihilation and the resultant products. ic effect and explain the Explain an Describe process. Describe Compton scattering and explain the result produets ofthis process Describe pair production and explain the result products of this process Discuss the production of characteristic rays Deseribe the process ofthe production of Auger electrons ‘Write the general form of the attenuation equation for mma photons alculate the reduction of gamma radiation using the general atenvation equation State the relationship between the linear attenuation coeffciens and the hallvalue lay: shotoelect el 3940. Nuclear Medicine and PET/CT Technology and Techniques Figure 2 Prysics is the study both of matter and energy and of the properties, forces, and interactions that influence the behavior of matter. Nuclear medicine applies the principles of physics to all aspects of radioactive decay, to ehe imerac- ‘don bewween radiation and matter, and tothe detection and measurement of quantities of radiation and radiation pro tection. A basic understanding of these principles is critical to the use of radioactive materials and radiation-detecting instrumentation. ELECTROMAGNETIC RADIATION RR Heat waves, radio waves, infrared light, visible light, ultra- violet light, and sctays and gamma rays are all forms of electromagnetic radiation (Figure 2-1). They elffer only in frequency and wavelength. Longer-wavelength, lower- frequency waves (heat and radio) have fess energy than the shorter-wavelength, higher-tequency waves (isible light serays, and gamma rays). The wave properties of light were first shown by Christian Huygens in 1678 in his experi- rents with the separation of light into the color spectrum. The particulate characteristics of light were not appreciated tuntl the experiments and research of Einstein, Planck, and Milliken in the early 1900s. Although electromagnetic energy has no mass, at very high frequencies it behaves more 45 a particle does, whereas at lower frequencies it behaves more as a wave does. The best way 10 think of electromagnetic radiation isas a wave packet called a photon. Electromagnetic spectrum Photons are chargeless bundles of energy that travel in a ‘vacuum at the velocity of light, , which is 3 x 10"em/sec 1 186,000 milesisec* ‘The wave nature of electromagnetic radiation is syrmbol- ized by the Greek letter lambda, 4. Note that the Greek lambda used to refer to electromagnetic radiation should not be mistaken for the radioactive decay constant, which is discussed later in this chapter. Electromagnetic waves, as, their name indicates, consist of fluctuating fields of electric and magnetic energy. Figure 2-2 shows the pattern of the wave cycle, Since light travels a. a constant velocicy, the oscillating electromagnetic field wavelength and frequency are related by the equation: Wy where ‘is the wavelength, vis the frequency, and ¢ is the velocity of light. The time for one wave period, or cycle, is teasured in cycles per second, called hertz Planck described the relationship of the electromagnetic wave frequency t0 the energy as E=hv where v represents the frequency of the electromagneaic wave and his Planck's constant, 6.625 x 10“erg-secicycles This equation can be manipulated to relate to energy and wavelength as‘This equation expresses the energy in electron volts (eV) to the wavelength measured in angstroms (10"ctn). The rela- tlonship indicates that photons with very short wavelengths thave high energy and vice versa, ATOMS AND MOLECULES Mater is anything that occupies space and has mass Ancient Greek philosophers theorized that all matter was composed of small indivisible pieces, which they called “atoms. An ators the smallest quantity of an element (€., bydrogen, carbon, oxygen) that retains all the chemical properties of that element, Atoms cannot ‘be broken into smaller particles without losing their chemical properties are classified by their characteristics of weight, number of subparucles, and chemical properties. Two or more ‘oms may combine to form 2 molecule, A molecule is the lest particle of a chemical compound that retains all ‘chemical characteristics of that compound. Molecules have as few a5 two or as many as hundreds of atoms; elore tens of thousands of different chernical com- mands can be created by changing the number of atoms the configuration of atoms within the molecule ‘The atom is made up of rwo basic parts, the nucleus and -oibital electrons. A simpte representation of an. atom is structure similar to a miniature solar system (Figure 2-3). is clementary model of an atom divides the atom into portions: nuclear (inthe center) and extranuclear (the sanding area). Three principal types of subatomic par- ‘compose an atom—protons, neutrons, and electrons, nucleus is composed of two types of these particles— and neutrons; hence protons and neutrons are ied nucleons. The nucleus is a cluster of these particles gives the atom most ofits mass ‘he extranuclear region of the atom is the area outside ‘cleus and is mostly empty space with electrons that rit the muclens like planets revolving around the sun Tegion of the atom, specifcally the outermost elec- {s responsible forall chemical intetactions with other sand isthe area in which most of the interactions of ation and matter occur. In 1960 the International Unions of Pure and Applied ics and Chemistry set the standard substance physical Physics of Nuclear Medicine 4 Velocity 1 ons (€} = Elche fils (soi in) 5 ons IH) = Moanetic fed (dated ine) Figure 22 Component energy fields of electromagnetic wave 8 “rte, Figure 23 Bohs atomic model with central nucleus surrounded by exzranuclear region scale tothe carbon-32 (°C) atom, whose mass is defined to bbe exactly 12.0000 atornie mass waits (arn). All other atomic and particle weights are measured with reference (0 the "C atom. Basic discussions of the composition of atomsare limited othe fundamental particles: electrons that orbit the nucleus and the panicles in the nucleus (protons and neutrons). Particle physics describes several families that hold many dozens of subatomic particles. For example, protons and neutrons are in the family called hadrons; electrons and neutrinos are in a family of “Light’-mass particles called leptons; and photons are in a group called bosons. In 1964 Murray Gell-Mann and George Zweig proposed that the hundreds of particles known at that time might be ‘composed of combinations of simplex fondamental particles that they called quarks (Figure 2-4), Experiments since that time have shown that there ate actually six quarks that compose substomic particles. Fortunately for the purposes of this book, @ descrip tion of the composition of the atom and the processes of radioactive decay using the most common particles can be used42 Nuclear Medicine and PET/CT Technology and Techniques Nucleus & 107"cm Proton Q 10cm ° <10 cm of the components within an ator, Electrons Electrons are the smallest of the subatomic particles and are foand in the extramuclear region of the atom. Electrons are also called negatrons and are given the symbol ¢ or They have one negative unit of charge (1.6 x 10° cou lombs) and a small mass of 9.1 x 10g, or 0.000549amu, and travel at about one tenth the velocity of light. Since electrons carry a negative charge, they are dellected by electric or magnetic fields. Their mass allows them to have Kinetic energy that is proportional to the square of their velocity. Electrons are held in their orbits around the nucleus by binding energy, which in conjunction with theit motion and centrifugal force keeps them from being antracted into the positively charged nucleus, An electrically neutral atom has an equal number of protons and electrons. Protons Protons are found within the nucleus of an atom and are symbolized by th letter p or p*. They have one positive unit of charge, which is equal but opposite to the charge of the electron, The protons within the nucleus provide its posi tive change. The proton has a mass of L67 x 0g, ot 1.00759amu, which is 1835 times that of an electron. The total number of protons in an atom is the atomic number, symbolized by the letter Z, and is unique for each element for example, one proton is hydrogen, two protons is helium, three protons is lithium, Neutrons, A neutral particle within the nucleus of the atom hal been theorized by Rutherford in 1920, but neutrons were not experimentally found until 1932 by Chadwick, Neutrons are slightly heavier than protons, have no electric charge, and have a mass of 1.00898ama.. For simplification, the ‘mass of a neutron and a proton are considered to be the same—Lamu., Neutrons are symbolized by the leter n For this discussion a neutron can be considered a combination of a proton and an electron, Neutrons are unstable particles and break down into the simpler, more stable particles of a proton, an electron (beta-minus particle), and 2 neutrino. This instability of the neutron is the source of ‘one type of radioactive decay. The decay process and the resultant particles are discussed later. ATOMIC STRUCTURE 7 ‘The model of the atom described by Niels Bohr in 1913 is one of the most easily understood representations. This miniature solar system has mostly empty space with the electrons in orbit around a small central nucleus, Three subatomic particles are used in this model. The electrons ‘occupy specific orbits, or shells, around the centrally located nucleus. The electron shells are labeled K, L, M,N, ©, P and Q (see Figure 2-3), beginning with the innermost K shell. Specific amounts of energy hold each electron in its orbit. The innermost electrons are more tightly bound to the nucleus, and therefore more energy is required to remove them. Outer electrons aze more loosely bound and only require smaller amounts of energy in order for them to be removed. Electrons can be removed from their orbits only by overcoming the binding energy for that shel. Binding energies are greatest for the innermost electrons, and the binding energy of specific electron shells is greater in heavier elements The electrons do not actually revolve around the nucleus in circular orbits in one plane; they move around the nucleus in a spherical pattern, Collectively, the electrons swarming about the nucleus form an electron cloud. Indi vidually the electrons change their distance from the rucleus and can even occasionally pass right through the rmucleus. Remember that the chemical properties of the atom are determined by the outermost electrons. The number of cleetrons that can occupy each orbit is limited. The formula 2n’ defines the number of electrons that can be contained in the majorshells, where nisthe shell number. For example, the K shell can contain only two electrons, the second shell ‘can contain no more than eight, the thitd shell may contain 18, the fourth shell may contain 32, and so on. Each major shell is composed of several subshells ‘The nucleus of an atom has an approximate diameter of 10cm (See Figure 2-3) and is composed of a cluster of° = 100 150 Neuron number Figure 25 Neutron-proton ratio with line of nuclear stability. protons and neutrons. Most of the matter in the atom is located here; therefore the density ofthe nucleusisextremely ‘high. Particles in the nucleus—protons and neutrons—are gnown as nucleons, The simplest atom, hydrogen, consists one proton, no neutrons, an one orbital electron. The “second element, helium, has a nucleus consisting of two rotons, two neutrons, and two orbital electrons. More plex atoms have an increased mumber of protons and rons. In approximately the first 20 elements, the ratio ‘As with electrons in the extranuclear orbit structure of atom, the protons and neutrons in the nucleus are bound there with a specific amount of binding energy. bility ofthe nucleus is determined by the total binding nergy of nucleons in addition to the number and configu- ation of the proton-neutron arrangement, Protons and tuons tend to pair up, creating @ more stable nucleus, pending on the neutron-to-proton ratio. Nucleons move about within the nucleus and occupy certain energy states. ome nuclei hold additional energy in a nearly stable, oF melastable, sate. The nucleus can emit this extta energy in iianner called an isomeric transition, with the emission of electromagnetic radiation (gamma ray). MASS-ENERGY RELATIONSHIP S081 Einstein's equation E = me, c represents the velocity of i measured in centimeters per second, m is mass mea- ated in grams, and E is energy measured in ergs. Accord- ing to this equation, matter can be converted into energy, 1d conversely, energy can create matter. This equation pplis oa mass that is not moving and is therefore termed rest mass of a particle of matter. When matter is con- ted to energy, the type of energy produced is of a form that isnot imparted to or held by matter, such as heat ot Physics of Nuclear Medicine 43 binding energy. Rather, i is pure energy, electromagnetic radiation, ‘The test mass of an electron, 9.1 x 10g, can be found. by this equation to be equivalent to 0.511 MeV. Similarly, the rest mass of a proton can be found to be 931 MeV, and the rest mass of @ neutron 939MeV. This relationship is further defined in the discussions of radioactive decay and the interaction of radiation and matter. Mass Defect ‘The relationship between mass and energy can be observed in the strong nuclear forces that exist in the nucleus. The suum of the masses of the nucleons in a "°C atom is defined and measured as 12.00000amu. However, the mass of protons 1.00759 aru; the mass ofaneutronis 1.00898 amu; and the mass of an electron is 0.00054amu, There are six protons, six neutrons, and six electrons in this atom, given the total mass of the individual particles as 12.20266amu Since the carbon atom weighs 12.00000amu, a difference of 0.10266amu has been converted into binding energy of the particles in the mucleus. The difference in mass of the constituent particles and the total mass is called the mass defect. This amount of mass converts to 95.6279 MeV of nuclear binding energy. Particles can be removed from the atom only by expending a force greater than the binding energy. Nuclear Stability ‘The number of protons and neutrons that form all possible configurations of the nucleus are graphed in Figure 2-5, Any configuration of protons and neutrons forming an atom iscalled a nuclide, Of the approximately 3100 nuclides, most are unstable and spontaneously release energy of subatomic particles in an attempt to reach a more stable state, This nuclear instability is the basts for the process called radioactive decay. Approximately 270 of the nuclides are in a stable form, comprising only 83 elements. The remainder ‘of the approximately 3100 nuclides are radioactive and are termed radionuclides The cause of nuclear instability stems from the energy configuration of protons and neutrons in the nucleus, Although the structure of the nucleus might be thought of as being like a cluster of grapes, each nucleon has a certain ‘energy siate, For one specific radionuclide there isa specific release of particles and a specific amount of energy released by its unstable atoms in this decay process. For example, I always decays by emitting beta-minus (B°) particles and gamma rays of specific energies. From radioactive decay, each radionuclide has its own “fingerprint” of characteristic radiation() In Figure 2-5 the approximately 270 stable nuclides follow along the center of the line of stability. with44 Nuclear Medicine and PET/CT Technology and Techniques B Rh99 | Rh 100 | Rh 101 | R02 |Rh 103 aghr|16d| 2h — (45a [32 yr|4.0yr | 2094 [57m | 100 eb75 boring epze fttisde’ fe le paar od oct bs , ri FAS 799 | Statens PINSHLI, fra 8 Ly Ru 98 | Ru 99 | Ru 100 | Ru 101 |Ru 102 o Bt 1.87 12.72 12.62 1707 JL6L os on O88 82 e123 97.9082 98.9059 999082 | 100.9053 |_ 101.9043 Te97,| Te9$ | Te99 | Te 100 | Te 101 Na) 2X10%y] 1.5X10°y Jenr Bx10yr| 17s idm lr |e BS. wy [629° | pasra26 | pri3ytor 096 775466 |y.140 459 | y 318413 Mo 94 | Mo95 | Mo 96 | Mo97 | Mo 98 | Mo 99 |Mo 100 9.04 15.72 16.53 9.46 23.78 Thr 9.63 om ot o2 ost piaas | 6: 174,95 93,9080 94.9088 953046 | 96.9060 97.9084 99.9074 Zz N——» Figare 2-6 Section fom the chart of nuclides NUCLEAR NOTATION AND radionuclide forms located on ether side ofthisline, Radio- — NOCTEAR FAMILIES = nuclides above the line of stability represent a region with a relative abundance of protons, termed protonerich (or neutron-poor) mucli, Radionuclides it chis region achieve nuclear stability primarily hy undergoing the decay pro- ce55es of positton emission or electron capture, Radiom- cides below the line of stability in the neutron-rich (or procar-poor) area undergo a radioactive decay process called beta decay in which a neutton becomes a proton fotlowed by the release of a beta particle from the nucleus Some radionuclides are in @ metastable form, These radionuclides carry a slight amount of excess nuclear ‘energy, which can be emitted as electromagnetic radiation to allow the atom to reach @ more stable form. The plotting of nuclides by their number of protons (ertical axis) and neutron number (horizontal axis) isthe basis for the chart of the muclides (Figure 2-6). This chart contains all possibie muctides, both stable and unstable radioactive forms. The chart is uselu! in identifying daughter decay products from a patticular radionuclide ot series of decays, Each nuclide is represented in a small square containing information such as the halflife, decay mecha nism, tagiation type, and amount of ‘energy being emitted There 15 a standard notation in the physica sciences for representing elements with one- or two-letter chemical name symbol; its taken primarily from the Greek names. of the elements. The generalized symbol form commonly used in nuclear medicine appears as aX where X is the chemical symbel, Z is the element's atomic ‘number (number of protons), and A represents the atomic ‘mass number ofthe atom (protons plus neutrons). Although the number of neutrons can be indicated as a trating sub- script number, tis usually sot writen because ft can be calculated by subtracting Z from A, As an example, caebon with 6 protons and 8 neuttons would be written as C Because an atom with 6 protons will always be a carbon. atom, writing a 6 and C is 2 duplication of information, therefore by convention, it may be writen as '{C or simply as *C. Although early scientists learned much about determining atomic weights and organizing the elements into the periodic chart, they did not immediately understand thatthere could be different atomic forms ofthe same element. The same element will always have a specific number of protons, as listed by the atomic number Z, and will have the same chemical properties, however, the number of new- ‘rons can differ. The term nuctde refers to any configuration ofthe stom. The Greek terms iso, meaning same, and topos, meaning place, indicate that the atomns have the same position on the periodic table of elements, The term isotope, though sometimes erroneously used interchangeably with nudkde, actually defines a specific element with different forms of that element, each containing different numbers pl neutrons. "Te, "Te, Tc, ™Te, and "Te are al isotopes Of element 43 (see Figure 2-6) and follow a horizontal line ‘onthe chart of the nuclides, Since the periodic chat of the elements is far too small te comcait atte different isotopes ‘feacn element, the chart of the nuctides is commonly used to list the 3100 isoropes of all the elements. It is relatively simmpe to identify the isotopes since the char is laid out in ‘patter following the line of stability lisced by the ntumber of protons and neutrons (see Figure 2-5), Isotopes for each clement are found as horizontal line, since the number of neurons is ploted horizontally. ‘Additional families of nucle) are isotones, isobars, and Isomers, isocones are atoms of different elements that have the same number of neutrons but varying numbers of “protons. The following are isotones: Mo, Te, "Ru, and ‘gall having, 56 neutrons and forming a vertical line on ‘the chart of the ouclides Gee Figure 2-6). The isotones are found in vertical columns on the chart, as the number of ‘protons is represented on a vertical axis. Igobars ate found ‘ona 45-degree angle running from the lower right to upper Tet. Isobars are nuclides chat have equal weights oF the “same mass number (protons plus neutrons). Examples of isobars are Rh, Ra, Fe, and **Mo; they are found on a “5edegree angle in Figure 2-6 Isomers ae atoms that have identical physical attributes, fat as the number of protons, neutrons, and electrons, however, they contain a different amount of nuclear energy. Jsoimeric forms of an atom are identified by putting ant m after the mass number A. The m means the azom is cur tin a metastable form and will emit gamma radiation Trotm the mucleus 0 achieve the more stable energy configu- tation. The most commonly used radionuclide in nuclear inedicine is an isomer—”"Te. Technetium-99 (TO is a rove stable form with a hallife of 2.13 x 10” years, Other sorts that have been used in nuclear medicine are "ny and "Sr. On the chart of the nuclides, isomers are desig stated by a vertical line within the square for a specific radionuclide. Another format for diagramming nuclide information is presented with a hexagonal box for each nuclide, as in Figure 2-7 (he Teilivear Chart), where vertically adjacent neighbors are isobars. Obhque neighbors from upper left to lower right represent isotones, and obliquely adjacent neighbors from lower left to upper right represent isotopes. Ikomers are identified with a vertical line through the hhexagon, Within each hexagon is a box that idencifes the Physics of Nuclear Medicine 45 element symbol, atomic number, and mass number, As at example of the information contained in this chart, Mo and Te should be carefully examined. In Figure 2-7 molybdentum-89 has a physical hal-life of 2.76 days and emits both beta and gamma sadietion during radioactive decay. The maximum energy in MeV is listed as 1.230 for beta decay (B, and gamma-ray emissions ate identified Vertical downward arrows show that 92% of the time molybdenum decays to the isomeric form af "Fe and 8% of che ceme directly to ™Te. The lelt side of the hexagon representing technetium shows that isomeric transition Giscussed later) will occur 100% of the time that °*Te decays, Also indicated in the left side is the gamma energy of 0.140 MeV (or 140keV), Technetium-99, as represented. fon the right half ofthe box, is also raioactive, with a half Iife of 2.13 x 10° years. Decay is by beta radiation to its daughter product ruthenturn-99, which isa stable element Note that the "Fe decays first to "Te before i€ decays 10 stable rutheniam, DECAY PROCESSES: a Alpha Decay Alpha particles (helium nuclei consisting of two protons and two neutrons) are radioactive decay products from tadionuchides having large, unstable mass and subsequent alpha decay. Using standard nuclear notation, the parent clement M with atomic number Z ane) mass number A decays by alpha (a) emission as; $M=EIN-ac+ energy to the daughter product symbolized by N. Because alpha particles have 42 charge and a large mass of 4amu, they are very damaging to biological systems and therefore have no role in diagnastic nucleat medicine. Beta Decay The concept of a neutron being composed ofa proton and electron is important in certain types of radioactive decay. ‘A nucleus that is neutron rich becomes stable through, the conversion of one of its neutrons by the reaction: napah +o+ energy ‘Asa result, a proton and an electron (B”) have been created from the neutron. The mass number of the new mucle is the same as that of the parent nuclei, because the masses of a newtron and proton are virwally the same; however, a different element is created. The formation of a different clement by a radioactive decay process s called transmuta- tion, or isobaric radioactive decay, The 8° particle that has been created in the nucleus through this decay process is elected. In this decay process an antineutrino (9) is created and carries away part ofthe energy of this reaction, The lavis of conservation of tomentuim and energy ate accourited for by this particle. Neutrinos (¥) and antinew- tinos have no electrical charge and a mass of almost zer9,O08 6. Se" -@5@ 8585 o8e Pe = aToe of eats pote neg) Figure 2-8 Generalized beta particle energy spectrum travel atthe velocity of light and are almost undetect- ‘The excess energy from the neutron is shared between the beta particle and antineutrino. This sharing of kinetic meray is not equal. Sometimes the electron receives more he energy, anc the neutrino receives less, oF vice versa 451 result, the energy of the beta particle varies in a con: uous energy spectrum with some maximum energy Bax.) that was available from the nucleus (Figure 2-8). This ucear decay process is termed beta-minus (B°) or simply decay. The parent element M with atomic number Z dimass number A decays by beta emission as: {Ma pAN4B"+Ftenergy he energy released by this process is shared as kinetic anton) energy by the beta particle and the antinewtrino, pexample of a beta-emitting radionuclide used in nuclear icine is "1, which decays as follows: l= Bikes 474 ene 4B + v + energy (positron decay) pre=mav-+ energy (election caprure) siicon decay or emission resilts when a proton is sed (0 4 neutxon, a positron (BY), a neutrino, and In nuclear notation the general form of positron {5 writen as: Bt +e = 1OSUMeV) + 70.51Me¥) m example of positon decay is: yorpty Physics of Nuclear Medicine 47 ost rit Annihilation event ' Figure 2-9 Annihilation ofa positron and a nepatron. The resultant pair of O.5L1-MeV gamma rays travel 180 degrees apart ‘An emitted positron loses kinetic energy through collisions ‘with sureounding matter, because itis then moving slowly enough to be attracted to an electron. The positron and nnegatron spiral in toward each other, forming a temporary “positronium” atom for a brief instant before the two partt- cles undergo annihilation. In the annihilation interaction the mass of each particle (positron and negatron) is converted into pure electromagnetic energy. Using Einsteins equation (E= mc’), the rest mass of each particle is converted into a 0.511-MeV photon; thus two J51L-MeV photons are created, These two 0.511-MeV photons travel in opposite directions to conserve momentum (Figure 2.9) Electron capture occurs when an oxbital electron travels in che proximity of the nucleus ard is eapeured and com- bined with a proton to form a neutron. Remember from the cloud model of the atom that electrons spend a very small amount of their time in the proximity of the nucleus. Elec- tron capture is generalized as: IMte = AN vteneray ‘An example is the decay of "5 Site ='Btety Electrons involved ip elecuron capture are usually those fiom the K or L shell. The probability for capture from another energy level is very low. A result of electron capture is the ionization ofthe atom with subsequent relocation of48 Nuclear Medicine and PET/CT Technology and Techniques an outer shell electron ta fill the vacancy created by the capture event. Proton-rich muclei can reach stability through either positron emission or electron capture. These ate always competing processes, however, the configuration of the nuclei and 2 high energy content of some nuclei increase the probability of positron emission. If the energy content is less, electron capture occurs. Some atoms can undergo her process with a certain probability A commonly used radionuclide in nuclear medicine ©Ga, undergoes electron capture. In addition to the changes, that take place within the nucleus, characteristic x-rays are emitted because of energy changes that result fom the electrons cascading down to fill vacant positions Gamma Decay Gamma-ray emission representsa mechanism foran excited nucleus to release energy. The release of energy as a gamma ) ray may be part of another decay process, such as alpha or B*, In addition, it isthe process for releasing energy from. a metastable nucleus Gatmma-ray emission usually occurs when there is more shan 100keV of excess energy in the excited nucleus. It should be remembered that gamma rays and x-rays have the same characteristies but are named based on thelr origin, gamma tays being emitted from the nucleus and x- rays from the electron shells. The ideal radionuclides for nuclear medicine are those that emit only garama rays without emitting particulate radiation, These radionuclides thereby provide gamma rays for imaging without increasing patient radiation exposure, With patticulate radioactive decay, the nucleus is most often lefe with additional energy, which is released promptly in the form of electromagnetic radiation, When a metastable nucleus is present, a significant amount of time passes between any previous radioactive decay (such as from °*Mo to Te) and further release of energy. The release from the metastable state is termed an isomeric transition. tn this transition the nucleus goes from a higher energy level to a lower energy level through emission of electromagnetic radiation (usually greater than 100 keV); this is sometimes termed gamma decay. The equation for an isomeric transition may be written as follows: Ma My ‘An example of an isomeric transition is: Petes Btety ‘An alternative process to gamma-ray emission in metastable nuclei is termed internal conversion, which is the transfer of ‘energy from the nucleus to an orbital electron, which is, then ejected from the atom (Figure 2-10), Internal conver sion reactions usually involve electrons from the K shell but ‘occasionally involve electrons from the Lor M shell. In this, process the ejected electron is called a conversion electron. This leaves an ionized atom, which follows the normal Ieee L M Figure 2-10 Internal conversion of L-shell election. Energy transfer from nucleus ejecting orbital electron is alternative to gamma-ray emission, Internal conversion electron will have Trinetic energy equal to gamma-ray energy minus binding energy process of reshuffling its electrons, resulting in the release of characteristic x-rays or Auger electrons. Metastable nuclei are the purest sources of gamtna rays for nuclear medicine imaging. The isomeric forms for these medically important radionuclides require that they be generator produced within the nuclear medicine laboratory. Technetium-99m has become the most important radiomu- clide in nuclear medicine tmaging SCHEMATICS OF RADIOACTIVE DECAY ‘The various decay processes of radionuclides can be represented diagrammatically to illustrate the relationship of individual processes that take place in radioactive disintegration. These diagrams show the relationship of the parent radionuclide to the daughter nuclide ‘The diagrammatic representation of a decay scheme is based on representing the atomic number on the horizontal axis and the energy of the nucleus on the vertical axis (Figure 211, top). The parent nucleus, being larger, contains mote energy and is represented as a horizontal line at the top of the diagram. Through the process of radioactive decay the resulting daughter nucleus has less energy and is positioned Tower on the diagram. Arrows are used to illustrate the emission of radiation, These arrows show the transition of the nucleus by alpha decay as a decrease in atomic number (2), shifting to the lf, indicating that the atomic number is decreased by 2 in the decay process, In Br decay there isan increase in Z by 1, with a corresponding shift to the right. With positron decay and electron capture there is a decrease in Z by -1. Gamma emission is shown as.a vertical down arrow, indicating the release of electto: magnetic radiation and resulting only in a decrease in nuclear energy without a change in atomic number. The energy levels of the patent and daughter nuclei are repreEnergy Atomic mumber (Z) ‘Type of radiation Placement of daughter relative to parent nuclide Alpha (0) a “2 Beta(B) Betas") lectron capture (EC) Gamma emission ()) gure -1] Axes (op) show directions of increasing energy atomic nummer in decay schemes. Directional placercent of ach daughter nucleus is relarive to parent in decay schemes Peres $M oughter aan Pigure 2-12. Hypothetical decay scheme vepresenting beta pile emission, senied as hotizontal lines. The parent radionuclide, 444, i5| shown in general form as the decay process for B- emission (Figure 2-12) to the daughter product 4M. Two different diagonal arrows in this diagram tepresent two different energies of beta particle decay. In one case the longer diaga~ nal ine represents the emission of a beta particle going Aectly o the daughter nucleus without any additional nergy release. This beta particle energy equals the differ ence between the parent and daughter energy levels. The Physics of Nuclear Medicine 49 a Ty 8 dove oN S004 mov Figure 213 Generalized decay scheme for "1 Eve T= 67 hous 2% \e 0.140 Mev ome Shows 2X 10! yeors ro Figure 2-14 Generslized decay scheme for "Mo, shorter arrow represents release of a beta particle of lesser energy to an inuetmediate stare with the prompt release of & gamma ray (vertical down arrow) to the daughter product One individual atom of the parent radionuclide can release its energy in this beta decay by following either path; therefore one of two beta particles with different energies can be seen in this transition, and only those beta particles from the lower energy transition are accompanied by a gamma ray Figure 2-13 shows a simplified decay scheme for "1. The actual decay process for "1 has several different beta energies, though only one is prominent (93%). With this bata particle there is a gamma transition releasing 0.364 MeV, Figure 2-14 shows the decay scheme for "Mo followed by the transitions of "Te and Te to Ru, Approximately 82% ofthe transition from Mo to "Te is by one particular beta energy. The radionuclide is then in the metastable form of Tc, which has a half-life of 6 hours. The isomeric transition of "Te is represented as a vertical line downward of a 0.140-MeV gamma ray to the daughter product Te. Technetiue-99 has a long half-life (2.2 * 10° years) and decays by beta decay to yield stable "Ru. Figure 2-15 shows the decay process for "1, which is, used in research, The electron capture process is followed by an internal conversion process. The resulting daughter of tellurium, wich the electron vacancy left by internal conversion, subsequently emits K-chavacterste x-rays50 Nuclear Medicine an PET/CT Technology and Techniques L ic Ere Figure 215 Generalized decay scheme for ™L uA Photons. See Mean energy (keV) Gamma2 38 33 ‘Abundance percentage Gamma3 21 186 Gamma 16 3002 Gammas 3935 The tadioactive decay schemes shown in Figures 2-11 2-13, 2-14, and 2-15 have been simplified. Some radionu- lides can’ emit several different gamma-ray energies. as illustrated in Table 2-1 for "Ga. Galliumn-67 decays by electron capture to "Zn; the zine nucleus then emits several energies of gamma photons. As indicated in Table 2-1 there isa gamma-ray abundance percentage for each "Ga atom that decays. For every 100 atoms, approximately 38% emit gamma rays at 93.3KeV, and so on. Decay schemes of this nature are helpful in identifying the most abundant peaks for selecting imaging windows. Two, three, or four peaks can be selected for imaging on some instruments, In certain decay schemes, more than one gata ray is emitted for each atom that decays. Figure 2-16 shows a radionuclide that yields more than one gamaa ray for each atom that disintegrates. Decay by bets! leaves the daughter nucleus, with energy that will be released by gamma 1 and gamma 2. Decay of beta 2 is fallowed only by gamma 2. Mathematics of Decay. Individual atoms andergo spontaneous transformations to release energy and form daughter nucle. There is no way of predicting when that transformation will occur for any tone specific atom. However, when a lange mumber of atoms are present, 2 certain probability of radioactive decay is obtamed and a mathematical average rate of decay can be determined. A sample that contains N radioactive atoms has on average a certain number of atoms decaying per unit time represented as AN/AC. This can be described mathematically as Figure 2-16 Decay by beta | is followed by emission of gamma \ apd gamma 2. Emission of higher-energy beta 2 is followed only by singe gemma emission, garmma 2. Some radionuclides yield several gamma rays for single disintegration, where 2 is the decay constant of the radionuclide. In this cstation the minus sign indicates that AN/A? is negative, or decreasing with time. The decay constant 2 therefore represents a probability or average percent of the atoms present that will decay in a certain time period. The units of A are time or time. A value such as 0.10 hour? means that in each hout 10% of the atoms undergo radioactive decay. Through calculus the above equation is manipulated 10 derive the number of atoms remaining at any specific time (= Noe where N, is the number of atoms that remain at any time ¢, and Ny is the number of radioactive atoms at the original time zero. The factor & represents the fraction of radioactive atoms that remain after time ¢ and is called the decay factor. The € represents Euler's number, the base of natural logarithms (2.718); it has been raised to the power ~2t. The decay factor (€9 is an exponentially decreasing funetion with time and can be determined using a calculator (see Chapter D. The decay factors for various time intervals can be calculated; decay factors for “Te ate given in Table 22. ‘The number of radioactive atoms (N) is, by the preceding definition (AN/M = -AN), proportional to the radioactivity “Derivation ofthe general decay equation AN, ‘a anPE Decay factor recalibration factor 1.000 1.000 05 oa 1059 0891 12 0794 1259 707 Lala 0.830 1987 0561 1782 9.500 2.000 0445, 2247 0297 2518 0354 2.824 0315 aie bast 3.558 0.250 4.000 This equation can therefore be written to apply 10 radioactivity cis called the general decay equation: AA Ikis frequently difficult to work with the decay factor 2; jsmuch more practical to use another parameter fy, known hallife value. The decay constant A and ty2 are related by the equation: 0.693 “where 0.693 isthe value of the natural logarithm of 2. This relationship can be deriveri from the general decay equation by inserting values forthe initial activity (Aa) and the activ~ uy atthe time (@ when the activity (A) is at 50% of the tivity at time zero, The decay equation may be repre ‘sented by substituting the relationship of half-life for the decay constant and can be written as: Ae Aye 2% here the decay factor (DF? equals the exponential portion of this equation De fraction of elapsed time relative to the halflife can be _generated and used for determining the decay factor for any radionuclide, sometimes called the universal decay table. The “aleulation of decay factors is reviewed in Chapter 1. Since the decay factor represents an exponential furtetion, it can ‘be plotted either linearly or semilogarithmically (Figure 2- 17). When reviewing this graphic representation, note that the number of atoms remaining is 036 at one halflife and “observe that the semilogarithmic plot (Figure 2-17, bottom) “is represented as a straight line where the slope is deter: ined ay the decay constant Physies of Nuclear Medicine 51 1.0 Fos = £50 3 y 25 oO 10203 4 5 Time (half-lives) 1.00 05 Frocon of ocvty remaining 025 010 1 2039 4 § Time (half-lives) Figure 2.27 Fraction of radioactivity remaining (ecay ccnstaei) as function of time is shown fora linear plot op) and fa semilogarithmic graph @Ptiom). Slope of semilogarithmic suaight De is determined by decay constar, RADIOACTIVITY UNITS _ Kadigctivity is quantitatively measured as the number of atoms that disintegrate per unit time, Two different units can be used to describe a quantity of radioactivity, the cure (C3 and the becquerel (Ba). The curie was named in honot of Marie Curie, an early pioneer inthe study of radioactiv ity. One curie is defined as the amount of any radioactive material that decays at a rate of 3.7 x 10” disintegrations per second (dps). This basic unit can be multiplied by standard mathematical scientific notation to yield larger and smaller multiples ofthe curie52 Nuclear Medicine and PET/CT Technology and Techniques kilocurie (ki eure (Ci rnillcarie (mCi smierocurie (uC) rnancurte (nCi pleocurie (pC) 10 x C1= 3.7 x WP dps 10° x C1 = 37 x 10%dps 10" C1= 3.7 10°ds 0° x Ci = 3.7 x 10%dos 3.7% 10'dps 3.7 10 dps The Systeme International (SI) defines the other unit of radioactivity, the becquerel (Ba). In 1894, Henri Becquerel became the first person co identily radioactivity. One bece ‘quetel is the amount of radioactivity contained in a sample that decays ata rate of 1 dps. Because this unit is very small, nuclear medicine terminology uses multiples of the bec querel such as kilobecquerels (kBq) and megabecquerels (MBq). Because most radioactivity in nuclear medicine is admin: istered in doves of millicuries, or megabecquerels, 3 is important to be able to rapidly convert berween these two units of measure, Multiply the number of millicuries by 37 MBq/mCi to convert to the number of megabecquerels For example, 20mCi x 37 MBq/mCi= 740 MBa, Conversely convert the number of megabecquerels 10 millicuries by dividing by 37 MBeymCi, for example, 111 MBq/(37 MBa/ Simi ‘Vee quantities of radioactivity administered to most patients are on the order of several snllicuries (Several hundred MBq). The total radioactivity contained im a *Mo- Te radionuclide generator is on the order of many Ci (many tens of thousands of MBq). The quantity of radiac- tivity Contained in patient specitnens or assayed in a scintillation well counter is on the order of nanocuries t0 microcuties (kilobecquerels. Decay Calculations Decay factors (DF) other than those in decay tables (see Table 2-2) for "Te can be found by multiplying those that correspond to the decay interval. For example, a H4-hour DF equals the 12-hour DF (0.250) times the 2-hour DF (0.794), or 0.198, this result could also be obtained by multiplying the T-hour DF (0.445) by itself to get 0.198, or any other combination could be used. ‘Sometimes it is necessary to dispense ox wse a radiophar ‘maceutical prior to the calibration time. Precaliration DFs ray be calculated frost the inverse (I/DE) of the decay factor for the time interval. As an example, the precalibration decay factor for 2 hours equals 1/DF or 1/0.794 126. EXAMPLE I: A vial contains 10mCi of radioactivity remains after 2 hours? "Tc; how much DF for 2 hours = 0.794 1OmCI x 0.794 = 794m How much radioactivity was present 5 hours before for lomcr recalibration DF for 5 hours = 1.782 1OmCi x 1.782 = 17 82mCt How much of the 10m} remains after 36 hours? ‘DF for 12 haus is 0.250, which is applied three times: 1OmCi x 0.250 x 0.250 x 0250 = 0.156mCi EXAMPLE 2: Using the general decay equation, calculate the activity of SmCvml of TL aller 48hr Uy: = 73 hows) nee SmoCifnl x €?8 Set x eS #tehO In SaCi/ml xe SmCil x 0.630 317mCum| Average Half Life and Effective Half-Life Average halfife describes the average lifeime of an atom in a sample of radioactivity. Mathematically, he average hall- ‘fe can be calculated as, oe LEX tn Although this term represents the average interval for which a group of atoms exists, it has no clinical use in patient dose calculations. The decay constant of a radionuclide represents the physical radioactive decay. I does not indicate the rate of biological turnover. Because rates are additive, the biclogical decay constant can be added to the physical decay constant to give the effective decay constant: d= he ty Since 2 = 0.69342, dividing both sides of the equation by 10.693 can represent this equation in terms of the half: ‘where fy 36 the effective halblife (Figure 2-18), ¢ ts the biological half-ife, and tis the paysical halt-life, This equa tion may be reorganized into the following forme: (ox EM) Tine Figure 218 _Semilogarithmic plot of physical (,) and biological () hall-fe components of shorter elective halflife,cxinaple of effective half-life might be the determination the disappearance of "*"Te-MAA from the hungs. Assume that the biological hal-ife is 3 hours. ria iy Gho* Cho he effective allie Gu) inthe Tung is thus 2 hours Jnter Radionuclide Relationship Asatadionuctide decays to form another muctide,eransma- {ution takes place when there isa change from one element A another. Some radioactive decay relationships are critical othe success of muclear medicine, as these decay processes possible radionuclide generator systems that produce acive elements as daughter products from the decay ALhis point, the reader is referred to Chapter 6 and the ented Generator Systems. The discussion in this ion covers the equations that govern radionuclide gen- ling, the parent-daughter relationships that create lar and transient equilibrium systems. RACTIONS _— Electrically charged particles (alpha particles, electtons, 3 positrons) have a high probability of interacting with the matter through which they move. Their mass and elec- cal charges interact with the mass of the nucleus and ical charges of atoms. In addition, the kinetic energy ofthese particles and the properties of the surrounding Thatter determine how the particles will interact and how these particles will travel. The density of matter, its omic number, andl its mass number influence the type and probability of these interactions stable excitation ofan outer electron toa slightly higher gy level The outer electron is not removed from the |, and therefore no ionization oceuts. Typically, excita- Physics of Nuclear Meclicine 53 energy of an electron were 54keV and the incident energy were 7BkeV, the electron would be ejected from the atom with 24 keV Ge., TBKeV — 54keV) of kinetic energy, Ioniza- tion occurs in all forms of matter-—solids, liquids, and gases. Alpha Particles ‘An alpha (a) particle is equivalent to the nucleus of a helium atom ($He). Alpha particles are produced by radioactive decay in very large, unstable atoms. Their high mass and double-positive electrical charge give them a high ion- {zing potential and a short path length in solids, liquids, or gases. They typically have energies between 3 and 8MeV. Because it takes only about 34eV to create one ion pair, the high energy of alpha particles along with their high mass, and positive charge can create several hundred thousand ion paits in only a fraction of a millimeter: this can be devastating to biological systems. Radionuclides that emit alpha particles have no use in nuclear medicine Beta Particles ‘The interaction of negatively charged bera particles in the proximity of the nucleus of an atom results in them being attracted to the positively charged nucleus. As the beta particle is deflected and slowed in its path, there isa release of energy as x-rays, called bremsstrahlung radiation. These scrays are released in a continuous spectrum because of the variations in kinetic energy and path geometry of the beta particle. Bremsstrahlung interactions increase in probabil- iny with materials that have a high Z number. Abeta particle emitting radionuclides, such as *P, should not be shielded by a high-Z material such as lead. Its more appropriate to use a shielding material of low 2, such as plastic, because bremsstrahlung interactions are then less ‘Wkely co occur. Bremsstrahlung interaction is the method used 19 produce x-rays in computed tomography (CT) scanners, Annihilation Positively and negatively charged electrons (positrons and nnegatrons) are antiparticles of each other. When positrons, are produced in a decay process and are ejected from an atom, they have enough kinetic energy to travel a maximum, ‘of a few millimeters, The positron and an orbital electron from an atom are attracted to each other because of their ‘opposite charges, spiral in toward each other, and interact, in a process called annihilation. The path distance required fora positron to find an electton for this interaction depends ‘on the positron energy and number of electrons m the immediate area of the atom from which the positron was originally emitted. tn the annihilation process the mass of the two particles is canverted into energy according to Einstein's equation £ = mc’. The rest masses ofthe positron and negatron are identical (0.511 MeV), giving a total energy of 1,022 MeV. The mass of each particle is converted into a 0.511-MeV photon (see Figure 2-9), The two photons travel54 Nuclear Medicine and PETCT Technology and Techniques Interaction Interaction site Photoelectric Inner electron shells Compton (Outer electron shells Pair production Nuclear field Energy range Several keV t0.05MeV Several keV to several MeV <102MeV Secondary effects Photoelectron, characteristic stays Compton electron, scattered photon Positron, electron, annihilation photons in opposite directions, 180 + 0.5 degrees from each other, to conform to the law of the conservation of momentum. BY +e = 7OS1LMeV) + y(0.511 MeV) ‘The detection of these two annihilation photons is the foundation for positron emission tomography (PET) imaging, PHOTONS _— Electromagnetic radiation in the narrow range teferred to as visible light can be reflected or absorbed with little effect ‘on matter, However, atthe energies referred to as x-rays and gamma rays, the wavelength of the electromagnetic radia tion becomes smaller than the size of the atom and the radiation penetrates through matter. Electromagnetic radia tion, oF photons, is far more penetrating in matter than are particulate types of radiation. There is no specific range for these photons, and their interaction with matter is based ‘only on a probability of interaction, in matter, photons may undergo scattering, may have no interaction with matter, or, because they are simply energy, may be completely absorbed and disappear. Although photons are more penetrating than particulate radiation, they demonstrate absorption, which diminishes exponentially with the distance traveled, For photons with energies associated with x-rays. and gamma rays, three types of interactions occur: photoelectric, Compton, and pair production (Table 23), Photoelectric Effect ‘The photoelectric effect is an interaction that takes place between an incident photon and an inner orbital electron. For the photoelectric effect to occur, the energy of the incident photon must be greater than the binding energy of the orbital electron. In the photoelectric effect the photon energy is totally absorbed, with some of its energy used to break the bond of the electron in its shell and the remaining energy given tothe electron in the form of motion or kinetic energy (Figure 2-19), Therefore a generalized relationship for this interaction can be writen as: Photon energy = Electron binding energy + electron kinetic energy The photoelectric effect usually occurs with electrons found inthe K or L shell. Since an electron has been removed from, the atom, there is no longer an electrically neutral balance between the number of protons and electrons, therefore the atom has been ionized with an inner shell vacancy created, Figure 219 In photoelectric eect, incident photon is totally absorbed and transfers all is energy to resultant photoelectron, In this interaction an ion pair has been formed—the positively charged atom and the negatively charged photoelectron—that leaves the atom. ‘The electron vacancy can be filled (1) by another orbital electron dropping in to fll a vacancy with the subsequent emission ofa characteristic x-ray or (2) by an Auger electron (discussed later). The ejected photoelectron is no different from any other free electron in matter and is involved in other interactions, depending on its kinetic energy. The disappearance of the incident photon is important clinically because the energy has been absorbed completely by the patient ‘The probability of a photoelectric interaction occurring, depends on the energy of the incident gamma ray and the atomic numberof the material. Obviously, the photoelectric effect cannot occur unless the photon energy is above the binding energy. As a photon’s energy increases, the probability for photoelectric interactions decreases (Figure 2-20), The probability of 2 photoelectric interaction increases dramatically with the atomic number, that is, photoelectric interactions are unlikely to occur in low-Z materials such as water and tissue but are likely to occur in high-Z materials such as the iodine in a sodium iodide crystal or in lead, The photoelectric eflect is therefore the primary type of interaction for detecting gamma rays with nuclear medicine instruments. All the gammacray energy is given to the crystal material ina thallium-activated sodium iodide crystal, The photoelectric elfect is therefore one of the most important interactions for nuclear medicine applications.igure 220. Interactions by photoelectric, Compton, and pair seduction combine 10 form attenuation coefficient. Probability decreases rapidly with increasing photon energy erections through Compton scattering decrease more slowly, pair production becomes dominant above 1.02Me¥, tsorption of gamma rays ofa scattering of gamma radia The Compton effect involves an inelastic interaction I lower energy and longer wavelength. The energy and length of the scattered photon are always lewer than ose of the incident photon, and the scattered photon's td photon will beat a shallow seater ange and can be Glslte! using, the elatonship f Fw = F528, (O51) re Ey is the incident photon energy (in MeV). As an. example, fora 0.140-MeV gamma ray, Emn = 0.090 MeV: for (0.364-MeV gamma ray, Eng = O.150MeV, The maximum energy Enya. oF Backscatter energy, will be at a scatter angle of 180 degrees and can be caiculated #8 402555) is maximum scater energy for a 0.140-MeV photon is 049eV and for a 0.364-MeV photon is 0.214Me. The Physics of Nuclear Medicine 55 Figure 221 Comon scattering accucs in outer-shell clectrons, with scattered photons having lower energy and a Jmger wavelength. Ejection of Compton electron produces an ionized atom rmaxitnum scatter energy identifies the sharp increase in scatter on the gamma-ray spectrum, called the Compton edge. The likelihood of Compton scattering increases with atomic number, therefore there is more Compton scattering in materials with high atomic numbers. Compton interactions are less likely to occur with higher-energy photons (see Figure 2-20) Pair Production Pair production is an interaction produced when a photon with energy greater than 1.02 MeV passes near the high electric field ofthe nucleus, The steong electrical force brings about the energy-mass conversion. When the photon comes near the nucleus, it disappears totally and two particles of matter are created, an electron and a positron, each possess: ing the mass equivalence of 0.511 MeV. For this interaction to occur, the initial photon must possess 1.02MeV or more of energy. Any additional energy of the incident photon is converted into kinetic energy, which is given tothe positron ‘and negatron, thus conserving energy and momentum. The photon originally had zero charge, and che offsetting positive and negative charges of the negatron and positron ensure that the net charge remains zero, The fates of the negatron and the positron are the same as if those particles were created by radioactive decay processes. The negatron interacts with surrounding atoms, possibly causing ionization and excitation. The positron loses some of its energy through intetactions and ultimately undergoes annihilation with an orbital electron from an atom, producing a photon pair with 0.511 MeV each. Figure 2-22 illustrates the pair production process, Figure 2-20 shows the dominance of pair production interactions for photons with very high energy. Note chat56 Nuclear Medicine and PETICT Technology and Techniques edn photon >1.02Nav Figure 222 In pair production, photons with energy greater than 102MeV may interacr with strong forces near nucleus Positron-electron pair is created with kinetic energy equal 10 ‘excess above LOZMeV. Poston will uncergo annihilation with electron there can be no pait production interactions below 1.02MeV, EXTRANUCLEAR ENERGY RELEASE Si ‘Three distinct processes can occur in the electron structure of the atom to release energy. These result in che creation ‘of bremsstrahlung radiation, characteristic x-rays, or Auger electrons. Bremsstraflung Radiation Bremsstrahlung is a German word that simply means braking radiation, Bremsstrahlung is the process that results in the rapid deceleration of a charged particle as it comes under the intense positively charged electric field of the nucleus. The particle, whether itis an electron or a positron, has a very small mass in comparison with the nucleus. The rapidly moving particle is attracted toward or repulsed from the nucleus and, as it is rapidly decelerated, changes its direction. This deflection and deceleration results in a significant loss of energy, which is emitted in the form of electromagnetic radiation in the x-ray region (Figure 2-23) In this type of interaction the conservation of energy and ‘momentum must be maintained: therefore the energy'of the initial charged particle is equal to the surs of the energy of the particle after deflection and de emitted bremsstrahlung xray. The production of bremsstrahlung x-rays is greater for «toms with high Z values and can increase the total amount, (of radiation being produced and be more hazardous if thin lead shields are used, With thicker lead, the resulting bremsstrahlung emissions are attenuated within the shield The production of bremsstrahlung radiation is not as significant in materials with low atomic numbers, such as plastic; therefore itis better to shield pure beta-emitting radionuclides such as "P with plastic shielding instead of thin lead: regent slecron eamsohling then Figure 2-23 Deceleration of a charged particle passing near nucleus result in release of energy in the form of bremsstrahlung serays Figure 2-24 Characteristic xrays are produced when an electron fills a lower shell vacancy. Energy of an x-ray represents an energy difference between two electron shells (Characteristic X-Rays ‘The production of characteristic x-rays accurs in atoms that have electron vacancies in theit inner shells. As electrons drop down to Ail these inner shells from outer orbits, there is a release of electromagnetic energy in the x-ray region (Figure 2-24), The energy of characteristic x-rays is determined by the energy shell difference, electrons filling the K shell are more energetic than these that fll an L shell because of the proximity to the nucleus and the higher binding energy. Auger Electrons Characteristic x-rays are produced as part of the process of reducing excess energy when electrons fill vacancies in inner shells. An alternative to characteristic x-rays is the Auger (pronounced oh-zhay) effect. In this interaction the surplus energy is given to another orbital electron, which 1s ejected (Figure 2-25). The ejected electron is called an Auger electron, and the atom is now left with two vacancies occurring in the electron structure. These vacancies arefiled by electrons from outer orbits, followed by the gion of characteristic x-rays or secondary Auger elec- he production of Auger electrons resultsin incteased ation exposure when these processes take place within dy tissue ATION AND TRANSMISSION |OTONS ___ az th photoelectric, Compton, and pair production ssts, These interactions combine to forma value called Is the ptobability of attenuation per distance through an absovoer; therefore yz has the unit of le the initial intensity of radiation 1 is reduced by the il function ofthe linear auienvation coefficient tance traveled x, 10 obtain the value ofthe reduced easy ofthe radiation field L The linear atenuation coefficient pis related 10 the afvalue layer (HVL) of the material by 0.693 AVL Lis the thickness of absorber necessary to diminish intensity ofthe radiation to half its initial strengch general attenuation equation can be rewritten 10 sate the relationship to HVL 2s (2693/03), Assuming 0.9em of lead and an exposure of 5ttB/r, the following can be calculated using the 3 atemnation equation: Physics of Nuclear Medicine 57 Auger slacron Altemative process to production of characteristic cays is emission of Auger electron, As electron Sle lower energy energy is transl to release orbital electron, Atom (righ) is left with two vacancies, which wall be filled by outer orbical is with subsequent characteris x-rays or additonal Auger electron production 1 = stm) inde? *0* S4aRVheee™ BéeaR) Ve) 0.125 0.625eR VAD Since the thickness of the lead is equivalent to 3 HVLs, an altemative method to more simply solve this problem is to divide the intensity of the radiation field by 2, three times, or: AleSmRy/ehey/29/2) 2 s535¢mR yh) The measure of attenuation can also be caclalated using another parameter, the mass attenuation coefficient, tan which depends on the material's density. The units of this, coefficient are cm*/g, The relationship between these two attenuation coefficients is represented as: ane om density Using these attenuation relationships results in the fllow- ing expression: iehem) “The mass attenuation coefficient can be broken down into its components for the processes of photoelectric, Compton, ‘and pair production imeractions. ‘© Nuclear instability, the energy configuration of nucleons, is the cause of radioactive decay. © The notation used in denoting various nuclides is 2X, where X is the chemical symbol, Z is the element's atomic number (namber of protons), and A represents the atomic mass number of the element (protons plus euros)58 Nuclear Medicine and PET/CT Technology and Techniques * Radioactive decay mechanisms important to nuclear medicine and PET ate alpha decay, beta-minus decay, positron decay and electron capcure, and gemma decay. ‘+ Alpha particles (two protons and owo neutrons) are emitted only by very large cadionucle, and alpha- emitting radionuclides are not used for imaging or internal therapy ‘© Beta partices are high-speed electrons, they are associated with high local radiation exposure and are therefore used for therapeutic radiopharmaceuticals. Their decay may also be followed by gamma emissions Positron decay and electron capture are competing mechanisms for proton-rich radionuclides to lose energy. ‘ Isomeric decay is the loss of energy from the nucleus without a change in the number of nucleons (protons and neutrons), with the energy leaving as electromagnetic radiation, '* The general radioactive decay equation is A,= Ace or Ae ae # The units of radioactivity are curies (3.7 x 10!dps) ot becquerels (1 dps) *# To convert mCi to MBq, multiply mCi by 37 MBa/mG. *# Decay factors (DF) are calculated from the exponential portion of the general decay equation DF = ez and are used to compute tables of DF values. ‘The effective half-life (jp) describes both the physical decay and the biological turnover of a radionuclide and is calculated from the equation: where {is the biological half-life and f, is the physical halEife ‘The interaction of radiation and matter results primarily in the ionization of atoms ‘The interaction of gamma or x-tay photons and matter takes place by photoelectric, Compton, ot pait production interactions. Annihilation reactions are between positrons and electrons (antiparticles of one another). The rest mass of each particle is converted into electromagnetic radiation through Einstein’ equation E = mc?, Two annihilation photons are produced, each with an ‘energy of 511 keV, traveling 180 + 0.5 degrees from each other. The attenuation of high-energy photons follows the general attenuation equation | = fy" A material's hat-value layer is different for various energiesL, Stephen Graham, Jonathan M. rumentation CHAPTER OUTLINE RADIATION DETECTION, Gaeled Detectors Scintillation Detectors Spectrometry Liquid Scintillation Counting EMISSION COMPUTED TOMOGRAPHY Approaches to Camera Quality Single Photon Emission Computed Control Tomography Quality:control Phantoms Dedicated Cardiac Cameras Routine Camera Quality-Control Positron Emission Tomography Procedures Factors Affecting Count Rate SPECTICT Camera Systems Film-Processor Quality Control ANGER SCINTILLATION CAMERAS Effects of Resolution, Scatter, and Computer Quality Controt Coliimators ‘Attenuation in ECT SPECT System Quality Assurance Spatial Resolution and Sensitivity QUALITY CONTROL National Electrical Manufacturers Crystals Survey-Meter Quality Control ‘Association Standards Positioning Logic Dose-Calibrator Quality Control [MAXIMIZING IMAGE QUALITY Eneray Discrimination Quality Control of Nonimaging Image Quality and Signal-to-Nolse Camera Calibrations Scintillation Detectors Ratio “Image Formation Calibration of Multicrystal Well, Instrumentation Factors influencing “SOLID-STATE ANID PIXELLATED Counters SNR, Scintillation-Camera Quality Control gas-filled detector Geiger: Mueller detector intrinsic testing fonization chamber positron emission tomography (PET) pulse-height analyzer (PHA) scintillation camera scintillation detector single photon emission computed tomography (SPECT) testing photomultiplier tube (PMT) eld collimator dh at bal maximum (FWHM) probe system | Describe the construction and operating principles of gas-filled detectors, including ionization chambers, Geiger-Mueller detectors, and dose callibrators, Explain the operation of scintillation detectors and. photomultiplier tubes Describe the mechanisms to generate an energy spectrum with a scintillation detector. Diagram the differences between liquid scintillation outing systems and scintillation crystal systems Discuss countrate limitations of gas-filled and simulation detectors relative to dead time, efficiency, geometry, and attenuation Diagram an Anger design of a scintillation camera stem ancl explain the function of each component, Diagram and discuss the properties of various camera cnlimators, Discuss spatial resolution and sensitivity of the scintillation camera Diagram and describe the function of position logic circuitry for the scintillation camera, Desunbe the operation of a pulse-height analyzer for energy discrimination, proportional counter tomography well counter Explain how images are formed using photographic or dighal systems, Define SPECT and explain the principles behind emission computed tomography. Describe the relative advantages of multiple detector SPECT cameras. Describe the operation of positron imaging systems for dedicated PET tomography. Discuss SPECT/CT camera system design Describe quality-control procedures required for survey instruments, List the quality-control procedures required for the dose calibrator and how each procedure is performed and results interpreted. Explain quality-control requirements for nonimaging scintillation detectors List and discuss quality-control procedures required for the scinuillation camera system, including SPECT. Define NEMA standards and theit application to nuclear medicine Define signal-to-noise ratio and its influence om data ‘quality60 Nuclear Medicine and PET/CT Technology and Techniques Nuclear medicine is a high-technology discipline. Accord: ingly, much more emphasis is placed on the design, mant- facture, and use of instruments than in some other felds, The equipment evolves out of a multidiseiplinary approach, encompassing basic radiation physics, electonies, detector design, and computers, Although the nuclear medicine technologist does not need to bevome an expert in all of these fields, basic knowledge ofthe principles behind the equipment is vital to maximizing a given imaging systems performance. The technologist is the Key player, since he or she divectly interacts with the equipment. A technologist needs an understanding of how nuclear instrumemation works far beyond knowing “which buttons to push.” This chapter emphasizes the principles of radiation detection, image formation, and tomography. A knowledge of basic mathematics and radiation physics is assumed References to specific vendors or equipment are specifically avoided; the same general operating principles apply co all modern nuclear medicine imaging systems RADIATION DETECTION = ‘Two main types of detectors are used in nuclear medicine gas-filled detectors and scintillators. Gas-flled detectors are typically used in nonimaging instruments, whereas scintil- Iators form the basis of most imaging, instruments Gas-Filled Detectors The basic approach of a gas-filled detector 1s quite simple: Radiation is sensed by detecting the fonization of gas molecules produced by deposition of energy during passage of radiation through the gas-flled detector. In essence, a gas- filled detector is @ container of gas with (wo electtodes, one positive (the anode) and one negative (the cathode), When ionizing radiation produces ion paits in the gas, the resulting free electrons are attracied to the anode and the positively charged gas molecule ions are attracted to the cathode. (An ion pair is the positively charged gas molecule ion and the free electron that came from ic) This movernent ‘of charge produces an electrical signal from the detector Commonly used gases include helium, neon, argon, hy- rogen, vapors, and ait. Gases that have a. high affinity for electrons, such a5 oxygen or halogens, are not used, because these would compete with the anode for the free electrons. ‘Types of gas-filled detectors. There are three main types of gas-filled detectors: ionization chambers, proportional counters, and Geiger Mueller detectors. Although in practice many factors determine which type a given detector represents, in theory the type is given by the value of the applied voltage across the electrodes in the detector. The curve that relates the applied voltage to the signal from a gas-filled detector is shown in Figure 3-1. The y-axis is, labeled “number of ion pairs collected per event." An event refers to one alpha, beta, x-ray, ot gamma ray passing Ww jm ivy v | VE #ION PAIRS. COLLECTED. PER EVENT o 7500 APPLIED VOLTAGE Figure 3-1 Relationship between applied voltage snd number dof ion pairs collected per event in a gas-filled detector. The curve is divided into six regions: I, recombination; ‘nization; Il, proportional, IV, limited proportional, Mueller; and Vi, continuous or spontaneous discharge Geiger through the detector and depositing energy in the gas Gin the process of ionizing gas molecules). Because the electrodes have a voltage across them, the free electrons are attracted tothe anode and the gas molecule ionsare attracted to the cathode. This movement of charge produces an electrical signal across the electrodes with a size that is propor tional ro the numbers of electtons and gas ions (e., ion pairs) collected at the electrodes. In general, a gas-filled detector gives one electrical pulse for every alpha, beta, x ray, or gamma ray it detects, with the size of the pulse determined by the number of ion pairs collected at the electrodes. The curve in Figure 3-1 is divided into six regions: recombination, ionization, proportional, limited proportional, Geiger-Mueller, and continuous discharge These divisions imply different operational modes for gas-filled detector. When the applied voltage across the electrodes is very Tow, the gas ions and electrons feel very hittle attraction toward their respective electrodes. Even though many ion pairs may have been created in the gas by tonizing radiation, these gas ions andl electrons are more apt to recombine with each other than to be separated and collected at the clecirodes. Because there is reduced signal from a gas-filled detector in the recombination region of operation, detectors are not operated with voltages in this region, Ifthe applied voltage is increased to 100 to 400V, the electrodes apply sufficient force to overcome the mutual attraction of the gas molecule ions and free electrons. As a result, all the ion pairs produced by deposition of energy in the detector are collected. This means that the size of each pulse in this ionization region of operation is directly related to the amount of energy deposited in the gas by the jonization event, If the gas is air, ionization chambers can directly measure exposure in roentgens. ‘When the applied voltage is increased te 400 to 800, the fee electrons gain a significant amount of kinetic energy ton their way’ to the anode. The larger the voltage on the anode, the greater the electron’s acceleration and increase in kinetic energy. These electrons now have sufficient energy 10 cause ionization themselves, As they make their way to the anode, they produce more ion pairs. Mast of this additional ionization occurs near the anode, because by this

Nuclear Medicine and PET CT Technology and Techniques Nuclear Medicine and PET

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Nuclear Medicine and PET CT Technology and Techniques Nuclear Medicine and PET

Uploaded by

Copyright:

Available Formats

You might also like