You are on page 1of 28

1988 SCA Conference Paper Number 8812

NMR IMAGING:

PRINCIPLES AND RECENT PROGRESS by Leo D. Brown

GE NMR Instruments, Fremont, California

ABSTRACT
NMR imaging (MRI) is an increasingly important tool in clinical practice. Initial concerns over safety, siting, and economics have successfully been addressed as the use of MRI has increased in medical and non-medical settings. While MRI images may resemble x-ray CT images, the underlying principles and technology are quite different and in some respects complementary. For example, while CT is well suited to imaging solid rock, MRI shows only the fluids in rocks. Moreover, most of the parameters measured by MRI have no counterpart in x-ray images. An understanding of the basic physics and nomenclature of NMR is important in appreciating the advantages and limitations of this technology, in evaluating different imaging strategies, and in selecting appropriate instrumentation. The rotating frame of reference provides a simple and convenient method for illustrating basic terms and pulse sequences. The experimental parameters measured by MRI (spin density, longitudinal and transverse relaxation times, chemical shift, and flow effects) determine image appearance and relate directly or indirectly to parameters of interest to core analysts, such as porosity, oil content and water content. Recent advances in gradient technology have increased gradient strength while largely solving the troublesome problem of eddy currents. These improvements have important implications for improving signal-to-noise ratios and resolution. This can be illustrated both by means of test phantoms and on actual core samples. While proton imaging has rightly been prominent, the flexibility of NMR in detecting other nuclei such as sodium, carbon, fluorine, and deuterium promises to extend the utility of MRI. INTRODUCTION Nuclear magnetic resonance (NMR) has been studied for over four decades, and has been a tool in downhole logging and in analytical chemistry for over three decades. The last decade has seen the dramatic growth of NMR imaging or MRI (magnetic resonance imaging), as it is known in medical circles, from a laboratory experiment to a routine diagnostic procedure. Thousands of MRI studies are now performed daily, and over a thousand papers were presented on the subject at just one recent meeting (Society of Magnetic Resonance in Imaging meeting, August 1987). The rapid acceptance of the technique occurred in the face of concerns over cost, safety, and siting which were systematically addressed by manufacturers and users. This technology can be readily transferred to specialized instruments appropriate for whole core analysis and at considerably low cost. For example, a medical MR imager typically costs $2 million and requires that the general public be excluded within a radius of almost 40 feet from the magnet. A smaller instrument, optimized for core analysis, costs less than half as much and requires that the general public be excluded from only a 13 foot radius.

1988 SCA Conference Paper Number 8812

F i g u r e 1.

A 2 Tesla, 31 cm bore CSI imaging/spectroscopy system manufactured by General Electric NMR Instruments.

MRI AND CT
X-ray computed tomography (CT) may serve as a starting point for understanding MR imaging. Like CT, MRI produces 2-D images or slices of 3dimensional objects, analogous to slicing bread. Both CT and MRI perform this task via sophisticated computer manipulation of the raw data. CT and MR consoles may thus appear similar.

Operating Console
1

|ftadtanl (Ui ^

1
r
\
i ILJEJ&UJL i

I i i
i i

Keyboard ' Rxictkxi Button*

CRT "*N

n !
L

Gredtant Cols

l
*

'1

- i i J"l| , J iViayilet Mtftar

i i i

I
i

p
Compitar ....i

Bal

L_ > >

Rmfctr

|*|

1 J

Multi-Format Camara Magnate >_ ' lapaUnH

Figure 2.

Block diagram of an NMR imaging/spectroscopy system.

1988 SCA Conference Paper Number 8812

Radio frequency (RF) transmitters and receivers bear an analogous relation to x-ray tubes and detectors. Superficially, the CT gantry with its donut hole opening may resemble the MR magnet with its cylindrical shape. However, the magnet, which represents the heart of the MR imager, has no counterpart in CT. The magnet assembly contains the main coil, which produces a field of typically 2 Tesla or 20,000 gauss. For comparison, the earth's magnetic field is roughly 1/2 gauss. The magnet assembly also contains shim coils to improve the homogeneity (smoothness of the field), and gradient coils, which are required to spatially encode the data. While x-ray CT measures electron density and effective atomic number, MRI measures several parameters, most of which have no counterpart in x-ray techniques. The NMR parameters include proton density (i.e., the density of hydrogen atoms or alternatively the density of other nuclei as discussed below), T1 and T2 relaxation times, and chemical shift. Additional NMR parameters include flow, which can be highlighted in specialized imaging sequence. Some of these parameters relate directly to quantities of interest to core analysts. For example, from proton density and chemical shift data or from proton and carbon density, core porosity and oil and water saturations can easily be determined. Other quantities of interest, e.g. wetting state, can in principle be determined by NMR measurements, but not as directly. In acquiring data MRI has several distinct advantages over CT or classical techniques. Measurements are rapid, as in CT. However, CT images suffer artifacts caused by attenuation of the x-ray beam, and there is no analogous effect in MRI. MRI measurements may be made directly in any viewing plane (e.g. longitudinal, axial, oblique, etc.) and resolution in clay-free samples such as carbonates is typically much better than in CT. Whereas CT views both solids and liquids, only liquids or semi-solids are visible in MRI, with a wide range of contrast possible between liquids. Finally, as opposed to CT, MR imagers have very few moving parts, reducing maintenance and obsolescence problems. An MR imager can thus be kept state of the art by software advances and occasional exchange of electronic modules as new hardware technology develops. NMR FUNDAMENTALS The mechanism of signal generation in NMR imaging is totally different from x-ray techniques, and a basic knowledge of NMR principles is essential to appreciate the data obtained. NMR physics rests on a sound theoretical foundation in quantum mechanics. Fortunately, simple models allow a basic understanding of NMR without resort to complicated mathematics. In this simplified mode, protons are considered to be small spheres that spin. Spin is the quantum mechanical property that is responsible for bulk magnetic effects. In the absence of a magnetic field, the spins are oriented randomly. However, in the presence of a magnetic field the proton spins may align with or against the external field. Moreover, like a spinning top, the protons in the presence of a magnetic field precess about the axis defined by the applied field with a characteristic frequency v and with a phase 9 . In general the phases are random.

NUCLEAR SWN AND MAGNETIC MOMENT

Figure 3.

Protons with the quantum mechanical property of spin.

RANDOM ORENTATION OF NUCLEAR SPMS M THE ABSENCE OF A POLARGONG MAGNETIC FELD

Figure 4.

Spins are randomly oriented in the absence of a magnetic field.

0RENTAT10N0F SRN-% NUCLEAR MAGNETIC MOMENT M A STATIC MAGNETIC FELD

Figure 5.

In a magnetic field, protons may align with or against the field. Spin up is termed parallel; spin down, antiparallel.

SPACE QUANTIZATION OF SPIN-tt MAGNETIC MOMENT IN A STATIC MAGNETIC FIELD

Figure 6.

Proton s p i n s p r o c e s s l i k e a spinning t o p .

RANDOM PHASES OF AN ASSEMBLY OF PRECESSING MAGNETIC MOMENTS

c^y V
Figure 7. Parallel spins precessing with random phases. Normally more spins align with the field than against it, i.e. the lower energy state is preferred over the high energy state. The spins try to align with the magnetic field. Thermal energy, which is much greater than the magnetic energy, tends to equalize the spin alignment along and against the field. However, the magnetic energy does produce a tiny excess of spins that point along the field, and this net magnetization is what is observed by NMR. Since the phases are random, the sum of the individual proton magnetic vectors is a vector in the direction of the applied field (z direction).

Bo

MACROSCOPIC NUCLEAR MAGNETIZATION CAUSED BY A N ASSEMBLY O F SPIN-ft MAGNETIC MOMENTS

Figure 8.

Net magnetic vector from an assembly of spins.

TRANSITIONS INDUCED BY IRRADIATION AT THE RESONANCE FREQUENCY

Figure 9.

At equilibrium the population, pi, of the lower energy state is slightly greater than the population, p2, of the higher energy state. In resonance, energy AE (the difference between El and E2) is absorbed by the spin system, and the population of the higher energy state increases.

In the resonance condition (i.e. when energy at just the right frequency, the resonance frequency, is introduced) the spins can be induced to flip to the higher energy state. According to the Larmor equation, the resonance frequency is proportional to the external magnetic field. The proportionality constant,y , is called the gyromagnetic ratio and is a physical constant characteristic of each nucleus in the periodic table having non-zero spin. In the course of the NMR experiment the net magnetic vector may be tipped out of the vertical by 180 degrees, 90 degrees, or some fraction thereof.

90 NUTATION OF MAGNETIZATION MDUCED BY RESONANT RF RELD

Figure 10. A 90 degree RF pulse produces a B1 field which rotates the net
magnetization vector by 90 degrees into the transverse plane.

RESONANT RF HELD OF AMPLITUDE B,

Figure 11.

In the rotating frame model the precessional motion of the spins and of the B1 field is frozen. (as in Fig. 10) .

In viewing such flipping of spins a rotating frame model is employed. In this model the frame of reference rotates at frequency (v) like a merry-goround so that the precessional motion is frozen. The z axis is defined as the direction of the applied magnetic field and hence the direction of the original net magnetization vector of the proton spins. Radio frequency (RF) pulses rotate that net vector about the x or y axis, and all detection of the resulting RF signals occurs in the xy plane (Fig. 12). A signal is observed because the precessing magnetic moment induces a voltage in the receiver coil situated in the x-y (transverse) plane. When the net magnetization vector is tipped into the xy plane and detected, the signal obtained is called a free induction decay or FID. As the name suggests, the signal decays with time. (Fig. 13).

FREE PRECESSION OF TRANSVERSE MAGNETIZATION NU INDUCING A SIGNAL M THE RECOVER COIL

Mxy

Figure 12 .

A net magnetization vector in the transverse (xy) plane induces a signal in the receiver coil in the same plane.

1988 SCA Conference Paper Number 8812

CALCULATED FREE-INDUCING DECAY SIGNAL t -STOtfe V - i

I
*

Figure

13.

A simple f r e e i n d u c t i o n decay (FID)

signal.

"HOT" SPINS DISSIPATE EXCESS ENERGY TO "COOL" LATTICE

Spins (hot)

<M)
Figure 14. Excited (hot) spins lose energy to their environment, the lattice.

1988 SCA Conference Paper Number 8812

EVOLUTION OF THE LONGITUDINAL MAGNETIZATION

Ti
Figure 15.

2T,

3T,

With time the Mz component of the magnetization returns to its original equilibrium value following a 90 degree pulse.

90*
INVERSION-RECOVERY PULSE SEQUENCE

180*- T-90*

Figure 16.

The inversion recovery pulse sequence. The FID is observed immediately following the 90 degree pulse.

1988 SCA Conference Paper Number 8812

As the excited spins cool down (Fig. 14) and return to their original state, the net magnetization vector reestablishes itself in the z direction (Fig. 15). This is called spin-lattice or longitudinal relaxation. The exponential constant associated with the reestablishment of the original state is called T-. The spin-lattice relaxation phenomenon suggests a pulse sequence designed to highlight differences in T1 . The inversion recovery sequence consists of a 90 degree pulse followed by a 180 degree pulse after a time intervals (Fig. 16). The effect on the NMR signal of varying T can be easily seen in the rotating frame model. If, for example, x is very short the spins are tipped into the negative z direction and then into the xy plane for detection after very little T- relaxation has occurred. If T is long then considerable T- relaxation has occurred by the time the 90 degree pulse is applied and the signal observed. Thus an image obtained with an inversion recovery sequence is typically strongly T. weighted and its appearance can be manipulated by varying x There is a second kind of relaxation that can be measured. Immediately after resonance, the proton spins are in phase. As time passes, the spins dephase and return to their original random phases (Fig. 17). This is called spin-spin or transverse relaxation, and is governed by the relaxation time constant T2. The spin echo sequence is particularly sensitive to T 2 effects. This sequence consists of a 90 degree pulse followed after some time T by a 180 degree pulse. (Fig. 18).

DEPHASMQ OF THE TRANSVERSE MAGNETIZATION TO Tt* RELAXATION

Figure 17.

Dephasing of spins following a 90 degree pulse.

1988 SCA Conference Paper Number 8812

90* SPIN ECHO PULSE SEQUENCE

Jj/^Vv^

**A/\/|^

9Qr-T-180r 9Cf-T-18Cf
Figure 18.
The spin echo sequence. T2* is the effective time constant governing observed signal decay. T2* is always shorter than T2.

PMNOPLE OF SPM ECHO FORMATION

t-tT
Figure 19.
Formation of a spin echo. At time t=0 following pulse, the spins are in phase. At time t=x, the dephased spins are rotated by a 180 degree pulse fast precessing spins (F) catch up to the slowly spins (S) at t=2-c. a 90 degree partially around y. The precessing

The 180 degree pulse "refocuses" the spins so that an echo of the original signal (produced by the 90 degree pulse) is observed. The generation of such refocusing can be illustrated by a simple analogy to a foot race. The runners start together and gradually spread out or dephase during the course of the race. If however, the runners about face half way through the race (and if each runner runs at a constant speed), the runners will recross the starting line together (refocused), the faster runners having run farther than the slower runners. Spin phases that have spread out can similarly be refocused (Fig. 19). The actual signal observed is a function of (among other things) T and Tp. Thus spin echo images are typically strongly T2 weighted. It is important to note that whileT is an experimental variable, T1 and T2 reflect actual chemical properties of the liquid molecules -- their size, motion, bondedness, etc. (Fig. 20). It is possible by running several experiments to obtain individual maps of proton density, T1 and T2. There is still another NMR property which can be measured -- chemical shift. Chemical shift arises because the protons are shielded to some extent by their surrounding electrons from the external magnetic field. This shielding is only slightly different for protons in, for example, water as opposed to protons in oil. The difference in shielding is only on the order of a few parts per million (ppm), but can be detected by NMR. In fact, the existence of chemical shift is the basis of most NMR spectroscopy. Because the protons in water and oil experience slightly different fields due to electronic shielding, they resonate at slightly different frequencies. (Recall that Larmor's equation states the resonant frequency is directly proportional to the magnetic field). The result in FID from two types of protons is more complicated, but can be resolved into two frequency components by a process known as Fourier transformation (FT) (Fig. 21). We now have a NMR signal that reflects proton density, T1, and T2, and chemical shift, but as yet we have measured bulk properties rather than forming an image. Of course, bulk properties may be very useful, but images can be even more valuable, if only to indicate what portion of the sample is representative of the unaltered core. Recently published images of limestones, dolomites, clay-free Bentheim sandstone, and Pigpen sandstone show that features such as bedding planes, bugs, and even individual sand grains can be seen by MRI in a variety of rocks. (Edelstein, et al 1988; Vinegar, 1986; Rothwell and Vinegar, 1985) In contrast to CT, magnetic gradients are used in MRI to spatially encode the signal (Fig. 22). The sample is electronically divided into small volume elements called voxels, and each voxel is mapped onto a picture element (pixel) in the image (Fig. 23).

1988 SCA Conference Paper Number 8812

WMr

Chokwterol

Motecular W M M W
T

|Mt
* " *

Figure 20.

T1 and T 2 are dependent on molecular size and motion.

Frequency

M
TIME AND FREQUENCY DOMAIN NMR SIGNALS
Figure 21. Fourier transformation transforms the signal from the time to the frequency domain. A single frequency is associated with a simple FID (a). Two frequencies are associated with a more complicated FID (b).

1988 SCA Conference Paper Number 8812

LINEAR GRADIENT OF THE STATIC MAGNETIC FIELD

Figure 22.

A linear gradient applied along the x direction.

THE RELATIVE SIGNAL STRENGTH DETERMINES THE PIXEL VALUE

Figure 23.

Each picture element is associated with a different volume element (voxel).

1988 S C A Conference Paper Number 8812 Since frequency is proportional to field, the superposition of x,y, and z gradients onto the main field would allow us to sample each voxel one at a time or to assign each voxel a unique magnetic address. While this is possible in principle, it is inefficient, and more complicated imaging schemes are used. The result of the superposition of gradient fields on the sample produces a complicated FID (Fig. 24).

Figure 24.

(a) In the absence of gradient fields both samples resonate at the same frequency and produce a simple FID. (b) In the presence of a linear gradient Gx, the two samples experience different magnetic fields and accordingly resonate at different frequencies. This produces a more complicated FID, which can be Fourier transformed to reveal two different frequencies (See Fig. 21b).

Fourier transformation translates the FID from the time domain to the frequency domain, as in NMR spectroscopy (Fig. 21). The spatial information has been frequency encoded and can be sorted out by computer. Typically one gradient is used to select a slice to be imaged, a second gradient is phase encoded, a third gradient is frequency encoded, and a double Fourier transform is applied to sort out the spatial information. Unlike CT it is possible to observe multiple slices in the course of a single MRI experiment. Voxels may be cubic (isotropic) or not (anisotropic). Four-dimensional techniques, where chemical shift is the fourth dimension, are also possible. Because MRI is so flexible, a number of different techniques are available, both in terms of pulse sequences (e.g. inversion recovery or spin echo) and imaging schemes (e.g. 2-D or 3-D), which enhance or suppress different contrast mechanisms and produce images at varying speeds and with greater or lesser resolution. Thus while operating an NMR imager can be reduced to a push-button affair, intelligently guiding an efficient NMR study

1988 SCA Conference Paper Number 8812


requires real expertise. Fortunately, the widespread use of NMR in chemistry and medicine has produced a pool of qualified scientists. SHIELDED GRADIENT TECHNOLOGY Since gradients are essential to forming an MR image, recent progress at General Electric has focused on gradient technology as a key to improved images (Roemer et al, 1986; Mansfield and Chapman, 1986). GE's shielded gradients are designed to provide increased gradient strength while turning the gradients on and off (as occurs repeatedly during spatial encoding) in a crisper, cleaner, and squarer fashion. This allows the observation of shorter spin echoes with correspondingly greater signal to noise since less relaxation has occurred, and allows greater resolution, since usable signals can come from smaller voxels. Improved gradients also bring a general reduction in imaging and spectroscopic artifacts. Shielded gradients (Fig. 25) were designed to overcome the problem of eddy currents. Without shielding, flux lines from the gradients enter the cryostat and other parts of the magnet assembly (Fig. 26). Shielded gradients, however, concentrate the lines of flux on the sample, drastically reducing eddy current effects (Fig. 27). Figures 28, 29, and 30 illustrate how shielded gradients turn off much faster than conventional gradients even when high gradient strength is employed.

Figure 25.

S200 shielded gradient set, which produces 10 gauss/cm with 70 amps and a rise time of 120 microseconds with an applied 120 volts.

1988 SCA Conference Paper Number 8812


UNSHIELDED G? 2

15

OS

1.
OS
I

15

2 2 1 5 1 - OS 0 Z (METERS! 06 1 15

Figure 26.

Lines of constant flux for an unshielded z gradient

SHIELDED Gz

- 2 I - 2

1 - 15

1 - 1

I - 05

1 0 Z (METERS)

I OS

1 I

1 IS 2

Figure 27.

Lines of constant flux for a shielded z gradient.

1988 SCA Conference Paper Number 8812

UNSHIELDED Z GRADIENT
Eddy current decay from 3 G/cm

I
3 4 5

I l I l
6 7

Time (ms) after pulse turnoff

Figure 28.

Gradient settling (decay of field) following a square gradient pulse. Unshielded z gradient at 3 gauss/cm.

876-

UNCOMPENSATED SHIELDED GRADIENTS

I
c
Q) 0

5 |

S !a4-

lf i 2
u. 0-

if , z -J | l
0.1 0.3 0.5 0.7 0.9 1.1 time (ms) 1.3 1.5 1.7 1.9

Figure 4b

Figure 29.

Gradient setting. Shielded S150 gradients at 9 gauss/cm without wave form compensation. Note change in time scale.

1988 SCA Conference Paper Number 8812

E 4-

COMPENSATED SHIELDED GRADIENTS

a> <u

c c 2

i?>
fljTJ

fe.

w-vyvW'\/v\/WY/v~v~A-'Y^ AAsv-v J\fTj*rJw\\tf\lfv -i

0 . 1

0.3

0.5

0.7

0.9 1 . 1 time (ms)

1 . 3

1 . 5

17

19

Figure 30. Gradient settling as in Fig. 29 but with wave form compensation
to reduce settling time to approximately 300 milliseconds. The favorable results of this advance are illustrated in Figures 31 and 32. Fig. 31 shows an MRI test phantom consisting of a 5 mm NMR tube filled with doped water and capillary tubes. Figure 32 shows the phantom viewed end on, illustrating an in-plane (2-D) resolving power of 25 microns. Note, however, the slice thickness is relatively large (1 mm). Diffusion considerations limit NMR microscopy to two micron resolution at best (House, 1982), but such extremely high resolution, if at all obtainable, is likely to be limited to very small samples at high magnetic fields or to phantoms of considerable thickness. Sensitivity considerations limit practical MRI of cores (and most other objects) to voxels a fraction of a millimeter on a side.

1988 SCA Conference Paper Number 8812

Figure 31. NMR phantom consisting of a 5 mm NMR tube with seven capillary
pipet tubes which may be filled with doped water.

Figure 32.

End on view of NMR phantom showing 25 micron in-plane resolution. The image was obtained using a 32X256X256 data set for excitations, and a 6.4 mm field of view. One of the sixteen contiguous, 1 mm thick slices obtained is shown. Such high resolution is not required in or necessarily desirable for bore analysis due to practical requirements of sample size, time required, and limited sensitivity.

1988 SCA Conference Paper Number 8812


Figures 33-38 show a study of limestone core. The voxels in this study are 1/3 mm by 1/3 mm by 2/3 mm. One hundred slices of the core were obtained in 110 minutes using a very short echo time of 2.2 msec. This experiment illustrates both excellent resolution and the general trade-off of time for resolution. Images of lower resolution and of a single slice can be obtained in just a few seconds. Likewise proton spectra to determine porosity or oil/water saturation (Edelstein et al 1988; Timur 1969) may be acquired in only a few seconds.

Fig. 33

ttvci t j H I H I OH ft S J J

39 *5 Tt

VIUA* mm * ' 4 5 !

*****

i mm

t/ l i t
9r w # n

l*t
mm m

**Mi > t ft**C

t o t a l t t * | . 0 howe a $)<<* t n x k n t i f . ? M

Fig.

34

1988 SCA Conference Paper Number 8812

.t.li.

U S UlOM

rfWi"

' #

JuSa

:Fig.

35

Vtfj

Fig.

36

1988 SCA Conference Paper Number 8812

Fig .

-.> :m

./>

ig-

38

Figure 33-38.

Eighty-eight of more than one hundred contiguous 0.7 mm thick slices obtained in a 1.8 hour experiment. Data were collected using a short (2.2 millisecond) echo time and a 128x128x128 matrix with a 45 mm X 45 mm X 90 mm field of view. Sample courtesy of ARCO Oil and Gas Company of Piano, Texas.

1988 SCA Conference Paper Number 8812


While proton data will remain preeminent because of their naturally high sensitivity relative to other nuclei and relative abundance, other nuclei add to core analyst's arsenal of tools and show considerable promise. Figure 39 lists NMR characteristics of a few nuclei relevant to core analysis. Since water contains no 13C, this is an obvious way to separate oil and water signals. This is very important in shaly sandstones, where broadened proton signals are particularly challenging to imaging and spectroscopy. Sodium NMR offers an obvious way to measure salinity. Other nuclei, such as 2H and 19F, may be introduced artificially as tracers either in laboratory studies or directly in drilling mud to observe filtrate invasion. Fluorine is especially attractive because of its relatively high sensitivity and its easy addition to oil or water phases as a fluorinated hydrocarbon or as a fluorine containing salt, respectively (Edelstein, et al 1988).

MAGNETIC RESONANCE PROPERTIES OF SOME DIAGNOSTICALLY RELEVANT NUCLEI


B^-^LAy fc^,

RtL Equal Number Abundance ofNucWat Nucleus Constant FWd <*) 1 0040 1H 046x10-* cum 2H 0410 1.11 c 0430 100 % 0,000 100 *. 640 x T 4 011 *k

o n i i m j

HMO HI

attentate 4240 043 10.71 4046 1140 140

Figure 39.

Magnetic properties of selected nuclei. CONCLUSION

The accuracy of NMR for measuring core porosity and oil/water saturation has recently been shown to compare favorably with data obtained by slower means. The expanding and improving range of NMR techniques gives the expanding and improving range of NMR techniques gives the core analyst a powerful tool to measure the properties of fluids in rocks accurately, rapidly, non-destructively, and independently of other techniques which may be subject to different sources of error (Edelstein, et al 1988, Vinegar, et al 1989).

ACKNOWLEDGEMENTS The figures illustrating the principles of NMR were provided courtesy of GE Medical Systems and originally appeared in a booklet entitled "NMR, A Perspective on Imaging." Images in Figures 32-38 were acquired by Ralph Hurd of General Electric NMR .Instruments, Fremont, California using a CSI 2T/31 imaging/spectroscopy system equipped with Acustar (TM) gradients, as was the phantom image in Fig. 31. The author wishes to thank H.J. Vinegar and W.A. Edelstein for helpful comments on the manuscript.

SELECTED REFERENCES Edelstein, W.A. Vinegar, H.J., Tutunjian, P.N., Roemer, P.B., and Mueller, O.M., 1988, NMR Imaging in Core Analysis, SPE 18272, 63rd Ann. Tech. Conf., Society of Petroleum Engineers, Oct. 205, pp. 101-112. Mansfield, P. and Chapman, B., 1986, Active Magnetic Screening of Gradient Coils in NMR Imaging, Journal of Magnetic Resonance, v.66, pp. 573-576. House, W.V., 1982, Feasibility Study - NMR Imaging Microscope, NSF SBIR Final Report, NSF Publ. NSF/PCM - 81006. Roemer, P.B., Edelstein, W.A., and Hickey, J.S., 1986, Self-shielded Gradient Coils, Society of Magnetic Resonance in Medicine, 5th Ann. Mtg., p.1067. Rothwell, W.P. and Vinegar, H.J., 1985, Petrophysical Applications of NMR Imaging, Applied Optics, v. 24, pp. 3969-3972. Timur, A., 1969, Pulsed Nuclear Magnetic Resonance Studies of Porosity, Movable Fluid, and Permeability of Sandstones, Journal of Petroleum Technology, v. 246, pp 775-786. Vinegar, H.J., Tutorjian, P.N., Crabtree, P.T., Raffaldi, F.J., DiFoggio, R. , and Edelstein, W.A., 1989, NMR Spectroscopy of Tight, Gypsum - bearing Carbonates, submitted to SPWLA 30th Ann. Logging Symposium, June 11-14.

ABOUT THE AUTHOR Leo D. Brown received his B.A. from the University of Utah in Chemistry in 1970 and his M.A. and Ph.D. also in Chemistry from Harvard University in 1973 and 1977, respectively. After a post-doctoral fellowship at the Swiss Federal Institute of Technology (ETH) in Zurich, he has specialized in NMR and MRI marketing at Bruker Instruments Inc., IML Imaging Inc., and most recently GE NMR Instruments, where he has worked for the last three and a half years. He is a member of the Society of Core Analysts.