68
1, July 1982
over four times the efficiency of the prototype could be
constructed using state-of-the-art ring detectors.
LUK HU WE*UgI * ECU t1__ _ _
This work was supported by the National Cancer Institute
Grant#5 R01 CA32846. The authors wish to acknowledge
the work done by the Physics Instrument Shop in construc-
ting the tomograph and the secretarial assistance of Diane
Vecellio in preparation of the manuscript.
[1] D. E. Kuhl, and R. 0. Edwards, "Image separation radio-isotope scann-
ing," Radiology, Vol. 80, pp. 653-661, 1963.
[ 2] H. S. Winchell, R. M. Baldwin, and T. H. Lin, "Development of
1-123-labeled amines for brain studies: Localization of
1-123-iodophenylalkyl amines in rat brain," /. Nucl. Med., Vol. 21, pp.
940-946, 1980.
[ 3] H. S. Winchell, W. D. Horst, L. Braun, et al., "N-isopropyl-[123]
p-iodoamphetamine: Single pass brain uptake and washout; binding to
brain synaptosomes; and localization in dog and monkey brain," I.
Nucl. Med., Vol. 21, pp. 947-952, 1980.
[ 4] D. E. Kuhl, J. R. Barrio, S. C. Huang, et al., "Quantifying local cerebral
blood flow by N-isopropyl-p-[123] iodoamphetamine (IMP)
tomography," 1. Nucl. Med., Vol. 23, pp. 196-203, 1982.
[ 5] T. C. Hill, B. L. Holman, R. Lovett, et al., "Initial experience with SPECT
(single photon computerized tomography) of the brain using
Quantitative Evaluation of
Doppler Ultrasound
E. R. Greene, M. W. Eldridge, W. F. Voyles, F. G. Miranda, and J. G. Davis
Abstract
During the last two decades, various Doppler methods have been suc-
cessfully used to screen patients with significant cerebral and peripheral
vascular disease. In general terms, the principal advantages of Doppler
ultrasound techniques in the evaluation of atherosclerotic lesions are that
they: 1) are noninvasive, 2) are nontraumatic, 3) are relatively inexpensive,
4) provide anatomical and physiological data, and 5) provide direct and
dynamic measurements. Nevertheless, the general limitations of the techni-
ques are of equal importance: 1) the techniques are difficult in some sub-
jects due to obesity and anatomical variations; 2) the technique cannot ex-
amine tissues surrounded by air or bone; 3) the techniques require
operator skill and a thorough knowledge of human anatomy and cardio-
vascular dynamics; 4) the techniques have finite spatial resolutions which
may compromise the important measurement of vessel diameter, ulcera-
tion, and percent stenosis; and 5) the techniques have finite velocity
measuring capabilities which may compromise some measurements of
highly disturbed blood velocities outside the range of 2-200 cm/sec.
As clinical demands for the early diagnosis and quantification of vascular
lesions increased, improvements in Doppler ultrasonics and spectra
analysis significantly increased the technical and clinical capabilities of ex-
isting simple, inexpensive instruments. Presently, both anatomical and
physiological images along with quantitative Doppler spectra from super-
ficial and deep-lying vessels can be obtained. Consequently, the ability of
Vascular Laboratory, Clinical Research Division; Lovelace Medical Founda-
tion, and the University of New Mexico School of Medicine, Albuquerque,
New Mexico 87108
Manuscript received at IEEE April 30, 1982.
0278-0062/82/0700-0068$00.75 © 1982 IEEE
IEEE Transactions on Medical Imaging, Vol. MI-1, No.
N-isopropyl 1-123 p-iodoamphetamine: Concise communication," I
Nucl. Med. Vol. 23, pp. 191-195, 1982.
[6] K. A. Frey, D. M. Wieland, L. E. Brown, W. L. Rogers, and B. W. Agranoff,
"Development of a tomographic myelin scan, Ann. Neurol., Vol. 10,
pp. 214-221, 1981.
[7] T. F. Budinger, "Revival of clinical nuclear medicine brain imaging."J.
Nucl. Med., Vol. 22, pp. 1094-1097, 1981.
[81 R. E. Coleman, B. P. Drager, and R. J. Jaszczak, "Studying regional brain
function: A challenge for SPECT," J. Nucl. Med., Vol. 23, pp. 266-270,
1982.
[9] G. F. Knoll, and J. J. Williams, "Application of a ring pseudorandom
aperture for transverse section tomography," IEEE Trans. Nuc. Sci., Vol.
NS-245, pp. 581-586, 1977.
[10] J. J. Williams, "Design and investigation of a circular ring emission
tomograph," Doctoral Thesis, University of Michigan, 1979.
[11] W. P. Snapp, "Reconstruction methods for nuclear emission
tomography," Systems Engineering Laboratory Technical Report #SEL-
TR-144, Department of Electrical & Computer Engineering, University
of Michigan, July 1980.
[12] H. 0. Anger, "A scintillation camera with multichannel collimators," J.
Nucl. Med., Vol. 5, pp. 515-531, 1964.
[13] C. Burnham, J. Bradshaw, D. Kaufman, D. Chesler, and G. Brownell,
"One dimensional scintillation cameras for positron ECT ring
detectors," IEEE Trans. Nuc. Sci., Vol. NS-28, pp. 109-113, 1981.
[14] W. L. Rogers, and R. S. Adler, "Time-coded aperture design for nuclear
medicine imaging: A study of signal-to-noise ratio," Applied Optics,
Vol. 21, pp. 324-333, 1982.
[15] K. F. Koral, W. L. Rogers, and G. F. Knoll, "Digital tomographic imaging
with a time-modulated pseudorandom coded aperture and an Anger
camera," 1. Nucl. Med., Vol. 16, pp. 402-413, 1975.
Atherosclerosis Using
new expensive imaging equipment to quantitate atherosclerotic lesions us-
ing spectral analysis techniques compares favorably with the interpreta-
tional precision of standard invasive or intravenous digital angiography.
New data suggest that unique hemodynamic information which reflects the
effects of cardiac output and vascular input impedance on the hemo-
dynamic consequences of an anatomical lesion can also be obtained.
This paper will 1) briefly discuss the general considerations of Doppler
ultrasonics; 2) critique the specific characteristics and utility of standard
clinical Doppler units; and 3) discuss the ability of new, multipurpose equip-
ment to quantitate (both anatomically and physiologically) atherosclerotic
lesions throughout the cardiovascular system.
*||tFUUMAIVv-
A quantitative evaluation of the effect of atherosclerosis
on the blood transport capabilities of the human cardiovas-
cular system would ideally require a method which could
noninvasively, nontraumatically, and dynamically measure
blood vessel morphology and intraluminal pressure and
flow variables.
Recent technological developments in ultrasonic echo-
Doppler instruments have resulted in systems which appear
more capable of exhibiting the characteristics of an ideal car-
diovascular diagnostic and research tool. Many of these new
technological capabilities have been sophisticated modifica-
Greene et al.: Quantitative Evaluation of Atherosclerosis
tions of the initial simple, inexpensive Doppler units [27, 2].
The impetus for the improvement of the technical capabili-
ties of Doppler instruments has resulted from 1) the un-
familiarity of clinicians with the physiological and dynamic
nature of the Doppler blood velocity, 2) the uncertainty in
medicine on the clinical importance of detecting multiple
grades of stenotic lesions, 3) the technical difficulties in pro-
ducing accurate and reproducible measurements of the
Doppler flow variables required to quantitatively correlate
disturbed flow regions with atherosclerotic lesions, and 4)
the requirement of reliable and precise measurements of
the variables needed to calculate volumetric flow (cm3/sec).
Moreover, we still do not fully understand the epidemiology
of vascular disease, the pathophysiology of occlusive
changes, the natural history of arterial lesions, or the precise
mechanisms of transient ischemic attacks distal to a stenotic
lesion. Doppler ultrasound instruments were developed in
the hope of providing techniques which would help us
understand these phenomena.
Clinical Doppler ultrasound systems which at present are
commercially available can be listed in order of their com-
plexity and cost: 1) directional and nondirectional portable
continuous wave (CW) devices, 2) continuous and pulsed
wave Doppler (PD) flow imaging scanners, and 3) real-time
ultrasonic sector scanners incorporating a range-gated puls-
ed Doppler identified as a duplex scanner (DS). These
various methods differ not only in initial cost but also in their
technical capabilities, limitations, versatility, and clinical
reliability. Importantly, each Doppler procedure of nonin-
vasively evaluating atherosclerotic lesions requires different
degrees of operator skill and clinical interpretation of the
raw data. Presently, it appears that no best test exists but a
combination of various Doppler techniques (direct or in-
direct interrogation of a given vessel) produces the most ef-
fective way of diagnosis [1].
The purposes of this paper are threefold: 1) to briefly
describe the general, theoretical and technical aspects of
current ultrasonic Doppler methods; 2) to discuss the impor-
tant clinical advantages, assumptions, accuracy, reliability,
and technical problems associated with each method; and
3) to suggest future research and development which could
improve the accuracy and precision of noninvasive, non-
ionizing Doppler ultrasound in the quantitative evaluation of
the anatomy, pathophysiology, and natural history of
atherosclerosis.
Methods
General Considerations
Accurate use of any Doppler ultrasound device whether
continuous, pulsed, flow imaging, or real-time duplex re-
quires an appreciation of the hemodynamic phenomenon
being measured as well as an understanding of the interac-
tion of acoustic energy with biological tissues and moving
blood particles. A brief summary will be presented here,
and the reader is referred to the literature for a more com-
prehensive treatment [2, 4, 13, 39, 45, 62].
The term ultrasound is used to describe mechanical vibra-
tions at frequencies above the limit of human audibility.
Most nontraumatic ultrasound operates between 1-20
69
megaHertz (MHz) and produces peak energy levels at less
than 5.0 W/cm2. The acoustic waves are generated when a
piezoelectric crystal, the transducer, vibrates at its resonant
frequency after the application of a controlled burst of elec-
tromagnetic energy to its surface. When the vibrating
transducer contacts the skin, two important changes occur
as the acoustic energy propagates through the tissue. The
energy will be absorbed to reduce its intensity at each point
along the beam, and it will also be reflected at points along
the beam where the acoustic impedance changes. The
acoustic impedance depends on the mechanical properties
of the biological tissue. If ultrasonic energy from the
transducer is delivered in short bursts of energy, or pulsed,
the same transducer can then be used to receive reflected
energy.
If the location of the interface between two tissues with
different values of acoustic impedance is stationary, the fre-
quency of the reflected wave will be approximately the
same as the incoming wave and can be received by the
same or separate transducers for dynamic structural infor-
mation, such as cardiac chamber [24] and blood vessel
dimensions [19, 20]. In contrast, a moving interface will
cause the reflected signal frequency to be frequency-shifted
by a magnitude which is proportional to the velocity of the
interface in the direction of the sound beam axis [2]. In the
simplest form, this relationship is given by the classic Dop-
pler equation: A f = 2 V f cos 0
C
where Af represents the Doppler shift expressed in kHz; V is
the velocity vector of the interface generally expressed in
m/sec; f is the frequency of the transmitted ultrasound; C is
the velocity of sound (1400 m/sec) of the tissue media sup-
porting the acoustic energy; and 0 is the Doppler incident
angle between the transmitted beam and the velocity vec-
tor. Clearly, if the values of f and C are assumed constant (a
reasonable assumption in most applications) and 0 can be
measured or held constant, Af will be proportional to V.
When the incident sound beam traverses a blood vessel
and its energy is not absorbed or reflected by significant
amounts of calcium [36], small amounts of energy are ab-
sorbed by each moving red cell. Since the dimensions of the
red cells are small compared to the wave length of the
acoustic energy, the reflected wave is radiated in all direc-
tions [26]. This phenomenon is called backscatter and allows
detection of acoustic signals with the receiving transducer
held at any angle with the backscattering particles. If the red
cell is moving relative to the transducer, a backscattered
Doppler shift can be recorded [2]. Since a distribution of
blood velocities is present in each segment of a blood ves-
sel [45], a spectrum of Doppler shift frequencies will be
developed [53]. These spectra can become quite dynamic
and complex in physiological states of blood flow and parti-
cularly within the region of atherosclerotic lesions [28, 33].
Analysis of the Doppler shift frequencies is further compli-
cated by the inherent pulsatile motion of the blood vessel
[21].
Even if the velocity distributions of the interrogated blood
vessel are well-behaved and not influenced by a focal lesion
and/or changes in distal impedances [64], other factors
70 IEEE Transactions on Medical Imaging, Vol. MI-1, No. 1, July 1982

which affect the Doppler spectra must be known in order to preciation of the complexities involved in making the
be certain that the Doppler spectrum genuinely reflects the measurements and processing the data [5, 29, 52]. One
various blood velocities within the lumen. These factors in- must recognize that V and Q are variables that cannot be
clude the following: 1) whether there is axial migration of directly obtained with a single noninvasive measurement of
the red blood cells which would weigh the Doppler spectra one physiological variable. Noninvasive Doppler methods
toward the higher velocities, 2) whether there is a uniform of measuring V require the following three distinct steps:1)
sound beam which is required for uniform insonation of the location of the vessel, 2) determination of the Doppler inci-
vessel, and 3) whether there is distortion of the Doppler dent angle, and 3) measurement of the spatial average Af,
signal due to nonlinearities and overall band-pass which represents V. To calculate Q, a measurement of the
characteristics of the device. lumen diameter is also required.
If the mean frequency shift (At) can be electronically ex- The third step, determination of the value of Af and hence
tracted from the Doppler spectrum generated from a uni- V, can be quite difficult. Technological improvements have
provided more control of the blood velocity sensing region
formly insonified velocity profile, the spatial average veloci-
ty(V) as a function of time can be measured [4]. Volumetric known as the sample volume. Consequently, specific loca-
blood flow (Q) can then be determined from an indepen- tions of blood velocities can be more uniformly insonated
dent ultrasonic measurement of lumen diameter (D) (assum- to provide Doppler spectra which accurately reflect the true
mean velocity or distribution of velocities within the vessel
ing a circular cross-section) by the equation: [4, 46]. From these spectra, appropriate signal processing
D2 can yield a more reliable display of a normal or disturbed
4 velocity profile or a more accurate calculation of Q.
The Doppler ultrasound methods of calculating V and Q A summary of important design parameters and opera-
are deceivingly simple. This has led some investigators to tional controls of either CW or PD units which influence the
use simple, commercially available, noninvasive Doppler extraction of Doppler audio spectra from the received
systems to arrive at quantitative measurements without ap- ultrasonic signal is given in Table 1.
Table 1
Relationship of Doppler Design Parameter and Operational
Controls with Measurement Characteristics
Design Parameters
or Operational Controls Relationship Measurement Characteristics
1. Transmitter frequency Direct Resolution, scattering power
(all units) attenuation
Inverse Near field longitudinal
dimension

2. Transducer diameter Direct Returned power, near field lateral

(all units) dimension
Inverse Resolution, signal/noise ratio
3. Transducer Q Value Direct Returned power
(all units)
Inverse Bandwidth, resolution
4. Pulse width (pulsed Direct Returned power
units)
Inverse Resolution
5. Gate interval Direct Returned power
(pulsed units)

Inverse Resolution
6. Pulse repetition fre- Direct Processed power, maximum measur-
quency (pulsed units) able frequency (Nyquist)
Inverse Range
Greene et a/.: Quantitative Evaluation of Atherosclerosis
In current clinical practice, only the abnormality, normali-
ty, and degree of abnormality of the Doppler spectral pat-
terns generated by a Doppler unit form the basis of the
noninvasive detection and quantification of atherosclerotic
lesions. In spite of their clear theoretical importance [64],
quantitative Doppler flow measurements (Q) in diseased
vessels are still of unknown clinical utility. These
measurements are subject to all the physiological and
technical assumptions outlined above.
Doppler Audio Signal Analysis
Although several methods of Doppler signal analysis have
been developed to extract a mean Af or to correlate Dop-
pler spectral characteristics created by disturbed flow
regions to anatomical lesions [9, 31, 57], the initial clinical
analysis of the signal is generally an audible interpretation by
the person performing the procedure. In the majority of
cases, this audible interpretation provides the operator with
reliable feedback information which governs the quality of
the noninvasive examination and subsequent permanent
recordings [58, 61]. Audio interpretation is obviously the
simplest and cheapest method of detecting arterial lesions.
Clearly, the observer must learn to recognize and differen-
tiate normal and abnormal velocity signals. This skill is ob-
tained with extensive training. Furthermore, the method is
relatively subjective, qualitative, and it does not allow for a
permanent record.
The most common technique used to analyze and display
the detected Doppler shift signal is the inexpensive zero-
crossing frequency meter [2, 4, 26]. This device produces a
voltage output which is proportional to the number of zero-
crossings that occur in the Doppler audio signal. In the
analysis of narrow band spectra with a good signal-to-noise
ratio (>20/1), this method provides an analog signal which
adequately reflects the mean frequency within the spectra
[43]. Furthermore, it is possible to depict forward and
reverse flow velocity on a common or on separate chan-
Fig. 1. FFT display of image-guided pulsed Doppler spectra obtained from quent calculated indices derived from the spectra are also
angiographically normal (0% DR); mild (<20%DR); moderate (20-49% subject to the gains and dynamic range settings of the input
DR); and severe (50-99%) internal carotid maximum diameter reduc- Doppler signal and the FFT display. Fig. 2 demonstrates two
tion (DR). Note the significant increases in peak frequencies and
spectral width associated with increased stenosis. Waveforms are spectral waveforms obtained from a normal internal carotid
ainverted to signify flow away from transducer.
71
Fig. 2 FFT spectral waveform obtained from a normal internal carotid
artery at different Doppler gain settings which are operator controll-
ed. Note the increased spectral width at a gain setting of 10 db com-
pared to 4db which may influence the interpretation of the spectra
and hence the quantification of the degrees of arterial stenosis. It is
also important to note that small degrees of spectral broadening
can also be present in normal, unstenotic flow regions such as
vessel branches. An additional waveform, the mode frequency,
represents the frequency of highest amplitude and may be used to
estimate if' and V.
nels. The significant disadvantages of this method, especially
when applied to disturbed flow regions near lesions, are as
follows [52]: 1) the method requires high signal-to-noise
ratios which are often difficult to achieve from the Doppler
signal, 2) the output is not always linear with wide band in-
put audio frequencies, 3) the output varies with the
amplitude of the input audio signal and threshold control
settings, and 4) no display of the spectra width (which is sen-
sitive to disturbed flow regions) is obtained.
As a result of these difficulties, several generations of spec-
tral analyzers have been developed. The most recent and
versatile units are the digital fast Fourier transform (FFT)
systems [34, 54]. These expensive devices can be obtained
as multipurpose units applied to the audio output of any
Doppler instrument or they can be incorporated into an in-
tegrated Doppler system. Generally, the FFT technique ac-
curately and reliably displays normal and abnormal Doppler
audio spectra and allows various spectral indices (yet to be
standardized) to be calculated and correlated to anatomical
lesions [9, 31, 40, 57]. Fig. 1 illustrates the FFT display (kHz)
of Doppler spectra obtained from angiographically normal,
mild, moderate, and severe internal carotid diameter reduc-
tions (DR) taken from our lab. Note the significant increases
in peak frequencies and spectral width associated with in-
creased stenosis. Peak frequencies are significantly depen-
dent on the location of the sample volume [53] and will be
proportional to the degree of stenosis only until the stenosis
becomes hemodynamically significant. It is important to
realize that gray scale or color-coded FFT devices and subse-
artery at different Doppler gain settings. An additional
72 IEEE Transactions on Medical Imaging, Vol. MI-1, No. 1, July 1982

calculation, the mode, displays the frequency with the angiography is reserved only for surgical candidates. In addi-
highest energy level at any time (an estimate of Af and V). tion, unique baseline physiological data (segmental
Even in this example of narrow band spectra associated with pressures, qualitative flows, and vascular reserve) of signifi-
normal flow patterns, note the variation in gray scale cant importance to patient management are also obtained
presentation of the spectra. These variations may influence in a cost-effective manner. Few limitations are present in the
the calculations of various correlative indices associated application of continuous wave units to peripheral vascular
with spectral width. The effect of this and other controls on disease. In advanced occlusive arterial disease, no Doppler
the audio spectra display is even more pronounced in high shifts may be generated due to lack of blood velocity which
energy, broad band spectra generated from disturbed flow will not create high enough Doppler shifts to pass the band-
regions. Clearly, if one is to quantitatively analyze Doppler pass output filters of the unit. In addition, Doppler determin-
spectra generated from flow velocity fields created by dif- ed cuff blood pressures may be artificially elevated (>200
ferent degrees of arterial narrowings, careful attention to mmHg) due to arterial calcification. Generally, these condi-
the control and display settings of the instruments is most tions are self-evident and do not compromise the accuracy
important. As with all applications of noninvasive diagnostic of the noninvasive diagnosis.
ultrasound, optimal instrument control settings will vary The genuine success of simple, nonimaging continuous
with application and patient habitus. wave units in peripheral vascular disease can be attributed
The preceding theoretical comments and operational to the advanced state of the occlusive arterial disease
suggestions have been general to all three categories of (hemodynamically significant) in most patients presenting
Doppler ultrasonic instruments regardless of clinical applica- with resting or intermittent claudication. Significant flow and
tion. Each modality has specific characteristics, capabilities, pressure reduction can be detected reliably and correlated
and limitations which will now be discussed. to stenotic lesions demonstrated angiographically. Reliable
detection of nonhemodynamically significant lesions (ar-
Critique and Discussion_ bitrarily defined as <50% maximum diameter reduction in
Continuous Wave Devices any angiographic plane) which do not significantly reduce
Simple, inexpensive continuous wave (CW) devices are flow and pressure clearly put unresolvable technical
certainly the most versatile and widely used Doppler in- demands on simple continuous wave devices in peripheral
struments in clinical practice [7, 44]. Initial systems were and other applications.
nondirectional but more recent versions were designed to Application in Cerebrovascular Disease
differentiate flow directionality which is often of clinical im- Although their relative importance remains controversial,
portance. The major technical deficiency of the CW units is nonhemodynamically significant but emboligenic lesions in
their lack of depth resolution. Consequently, if two distinct the extracranial cerebrovascular system have been implicat-
flow fields (vessels) are insonated by the relatively large ed in cerebral ischemia (transient ischemic attacks) and in-
sample volume (created by the three dimensional overlap farcts (strokes) (10, 29, 48). Consequently, attempts to
of the beam patterns of the transmitting and receiving crys- noninvasively detect hemodynamically and nonhemody-
tals), unspecific Doppler audio spectra will be developed. As namically significant stenosis using continuous wave units
an example, overlapping veins and arteries, a common ana-
were undertaken. Due to inherent physiological and ana-
tomical situation, can distort the CW Doppler spectra. Direc- tomical complexities of the cerebral circulation, simple in-
tional units have helped differentiate flow fields, but they direct periorbital Doppler techniques are not accurate
have not completely resolved this problem. Although no enough to detect all hemodynamically significant lesions
range resolution is possible, continuous wave units allow
and can detect very few nonhemodynamically significant le-
ultrasonic Doppler signals to be continuously sampled. Con- sions in the extracranial vasculature [16]. Generally, total oc-
sequently, no artificial distortion (aliasing) of the audio signal clusion of the internal carotid gives a positive test result (par-
is produced by high frequency shifts associated with ticularly with compression maneuvers) and thus allows this
elevated blood velocities and/or small Doppler angles [4]. simple indirect procedure to be a useful and commonly ap-
This distortion can be created by pulsed units which sample plied adjunct to more direct methods [1].
high flow rates. The distortion due to aliasing can mimic ab- Direct, continuous wave Doppler interrogation of the
normal spectra created by anatomical lesions [60]. carotid bifurcation using audio interpretation [61], zero-
Application in Peripheral Vascular Disease crossing detection [55] and FFT methods [8] have allowed
In spite of its technical limitations, the simple hand-held useful differentiation between angiographically determined
continuous wave device (with or without sophisticated spec- significant and nonsignificant lesions. One of the difficulties
tral analysis) has found wide acceptance in the diagnosis of with these nonimaging methods is the critical problem of
hemodynamically significant (flow and pressure reducing) le- the proper identification of the three major vessels (com-
sions in the peripheral arteries during rest and exercise mon, internal, and external carotids) in a small region
[7, 31, 44]. In this application, the procedure can be under- [11, 50]. Consequently, Doppler flow imaging systems with
taken by a technician without extensive training in the inter- and without audio spectra analysis were developed.
pretation of Doppler spectra. Unpublished results from our
laboratory with over 100 angiographic correlates support Doppler Flow Imaging
published results with an overall accuracy of 98% compared Continuous Wave Flow Imaging
with angiography. Thus, in our institution, peripheral In order to meet the need of proper vessel identification in
Greene et al.: Quantitative Evaluation of Atherosclerosis 73

the carotid system, Doppler flow imaging systems have

been developed. The Doppler flow image is created by a
video display whose spot patterns indicate a detected flow
velocity. To be displayed, this velocity (frequency shift) must
exceed a certain threshold value. The transducer is coupled
to a scanning arm which can be used to create a map of the
detected flow loci. Continuous wave Dopplers can be used
to drive the transducer [511 and create the ultrasonic
arteriogram whose image format has been more readily ac-
ceptable to image-oriented radiologists and clinicians. One
advantage (of undetermined significance) of the flow image
over standard echo imaging (B-scan) is that the true flow
lumen can be displayed in many planes. Unfortunately, the
flow lumen image may not coincide precisely with the ac-
tual vessel wall if there are calcified plaques or other lesions
in the interrogated vessel. The resolution of the flow image
(which requires several minutes to produce from a sta-
tionary patient) is dependent on the lateral width of the
acoustic beam pattern whose present resolution [62]
precludes the detection of plaques less than 1 mm. For the
same reason, thhe imaged flow lumen may also be several
millimeters larger than actual size. Although an image of the
carotid bifurcation is generated by the Doppler allowing
identification of the vessles, the quantification of the
atherosclerotic involvement generally depends on concur- ~
rent analysis of the Doppler audio spectra [9, 57]. CW Dop-
pIer imaging correlates with angiography have been promis-
ing for both nonhemodynamically significant and
hemodynamically significant lesions [9, 56, 411. In an at-
tempt to quantiate stenotic lesions, Spencer and Reid [571
obtained best fit regression lines relating percent stenosis by
angiography to percent stenosis by Doppler frequency I
ratios obtained from CW Doppler flow imaging. Unfor- .4 j

Fig. 4. Angiographic (top) and freeze frame DS images (bottom) with FFT
spectra of a normal carotid bifurcation with the sample volume (SV)
placed in the external carotid (EC) and later in the internal carotid
(IC) distal to the common carotid (CC). Note the high diastolic flow
velocity in the IC compared to the EC. In our experience, real-time
noninvasive images of the carotid bifurcation which are similar to
angiographic presentation are not easily obtained. Often the far
walls of the EC and IC are nonvisualized, hence calibration of the
spectra into V and Q is not always feasible.

tunately, their results showed significant scatter. Color-

coded CW Doppler imaging [47, 63] has also proven suc-
cessful in reliably separating occlusive, significant, and non-
significant lesions, but the more elaborate color technique
~ ~-. ~ has yet to be shown more clinically efficacious compared to
earlier black and white displays. Although CW flow imaging
systems have made important contributions in detection of
Cmm X _ _cerebrovascular disease, their main limitation, spatial resolu-
tion, has prompted the development of pulsed Doppler
Fig. 3. A freeze frame of a DS image with the sample volume (SV) in a nor- flow imaging units.
mal common carotid artery (CC). Note the narrow band spectra Pulsed Doppler Flow Mapping
(velocities) created by undistributed laminar velocity profiles. Due
to the ability of the DS to provide measurements of Af, Doppler
Comparatively, pulsed Doppler (PD) flow imaging units are
angle and vessel diameter, the FFT spectra can be calibrated in KHz,
velocity (cm/sec) or flow (Umin.).
more complex and expensive than the CW devices [38].
Due to their capacity to detect flow velocity from discrete
74 IEEE Transactions on Medical Imaging, Vol. MI-1, No. 1, July 1982
lesions using flow imaging will probably come from improv-
ed methods of Doppler spectral/blood velocity analyses.
This is presently a difficult matter due to the spatial distribu-
tion of various disturbed velocity patterns within and distal
to lesions and their convolution with the generally ill-
defined CW and PD sample volumes [60]. In spite of the
clinical success of static flow imaging instruments, their pro-
blems of 1) spatial resolution, 2) narrow fields of view and
versatility, 3) time required from image production, and 4)
relatively poor image quality were not overlooked. Predic-
tively, the development of sophisticated and quite expen-
sive systems which incorporate ongoing high resolution
real-time echo imaging technology [20] and pulsed Doppler
velocimetry was achieved [6]. This instrumentation was
denoted as duplex scanning.
Duplex Scanning
Available duplex scanning (DS) systems vary in their em-
phasis on imaging, image processing, pulsed Doppler
capabilities, and audio spectra analysis [49]. Units are being
applied in cardiology [3, 32] and in carotid [12, 14, 15, 17,
25] and femoral arteries [13]. In spite of the improved
resolution in vessel imaging [19], quantitative evaluation of
the lesion has often rested on the analysis of the range-
gated Doppler audio spectra. Quantitative peak blood
velocities calculated from calibrated FFT velocity spectra
[41] have been used. Fig. 3 demonstrates Doppler audio
spectra sampled from a normal common carotid artery ex-
amined by a 5 MHz DS in our laboratory. Note the narrow
band Doppler spectrum created by undisturbed velocity
profiles. Due to the ability of DS to provide measurements
of the Doppler angle and vessel diameter, estimates of
volume flow rate can be calculated [4, 32]. Doppler spectra
Fig. 5. Angiographic (top) and freeze frame images (bottom) of an abnor- from normal external and internal carotids are shown in Fig.
mal carotid bifurcation with abnormal FFT spectra created by a
moderate stenotic lesion in the IC. Note the imaged vessel calcifica- 4. The high diastolic flow due to the low impedance
tion denoted by the white arrow. Abbreviations are the same as in vascular bed supplied by the internal artery is clearly distinct
Fig. 4. The spectral waveform can be electronically inverted if
desired but is displayed here in the downward direction to from the reduced diastolic flow of the
high impedance bed
demonstrate flow away from the transducer. It is most important to of the external carotid. The shape of the spectral waveforms
note that calculations of if and V are unreliable in the distinct demonstrates an important discriminatory feature of the DS
regions of disturbed flow.

points in space independently and simultaneously (multi- SERIX_Xg t 'Ett<FFfPNOZY A NCWWASIE &F 0 jA A
gated), PD flow imaging units theoretically provide more cn- $EN $&.o4t y

specific flow images and Doppler spectra [3]. Compared -~

~'PA
*- '
;*
~~~~~~~~~M*a A te AN~
*ieft :
=

with CW imaging, increased operator skill is required with _S Ct 0 A: PN

PD devices due to the problem of finding the desired vessel

in three dimensions rather than two as is done with CW
Doppler techniques. Similar to CW imaging but, multiplane 0V | t; < t 0F'gf
views of the carotid bifurcation can be obtained. As with W

CW imaging methods, angiographic correlates using PD im- .4

ff ii0 t00
U
aging and Doppler spectral analysis are encouraging [37]
and their pitfalls can be delineated [11]. Generally, imaging :e
errors caused by calcification can be overcome by sono- PNAtff f
graphic spectral analysis. It is important to note that
nonoperable occlusive disease can be reliably separated
from operable, highly stenotic lesions. Due to the finite
spatial and velocity resolution of Doppler flow imaging in-
struments (either CW or PD), improvements in image reso- Fig. 6. Serial clinical results in our laboratory using different Doppler
lution (even at higher frequencies) seem unlikely. Conse- methods to identify angiographically demonstrated IC lesions of
quently, improved quantitative evaluation of atherosclerotic >20% diameter reduction and occlusions. See test for details.
Greene et al.: Quantitative Evaluation of Atherosclerosis
method. By analyzing the Doppler signal, vessel identity can
be further defined by its inherent flow patterns. In Fig. 4,
spectral data are displayed in kilohertz to emphasize that
the velocity and/or flow calibration requirements of angle
and diameter are not always reliable in internal and external
carotids in consecutive patients in our laboratory. This is
contrary to reports of other investigators [12, 40]. Fig. 5 il-
lustrates a carotid bifurcation with Doppler audio spectra
created by a moderate stenotic lesion (10-49% diameter
reduction) in the internal carotid artery. Note the imaged
vessel calcification which provides morphological informa-
tion along with the Doppler physiological data. As sug-
gested by Fell et al [25], implementation of the FFT spectra
analysis with the expensive DS has provided new, refined
capabilities of lesion categorization. These noninvasive
capabilities appear to stress the quantitative limits of biplane
angiographic interpretations. By careful separation of exter-
nal and internal blood velocity signals and by noting the lack
of diastolic flow velocity in the proximal common carotid
[17], total occlusions can be reliably differentiated for highly
stenotic lesions in the internal carotid. Other calculated
spectral indices obtained from the DS have been proposed
and prospectively studied with favorable angiographic cor-
relates in the internal carotid [40]. As previously demonstrat-
ed by Fig. 2, careful attention must be given to gain and
other control settings to obtain reliable results.
Serial clinical results using different Doppler methods
which demonstrated the ability of our laboratory to nonin-
vasively identify angiographically demonstrated lesions
which obstructed >20% of the internal carotid lumen in dif-
Fig. 7. A freeze frame DS image with the sample volume (SV) placed in a
normal right renal artery (RRA). Anatomical landmarks include a
cross-section of the superior mesenteric artery (SMA) and the ab-
dominal aorta (AA). Note the narrow band spectra with significant
diastolic flow associated with the low impedance renal bed. Vessel
diameter and Doppler angle can be estimated to allow calibration
in V and Q. These spectra patterns will change with the region of a
stenotic lesion. Significant increases in distal vascular impedance
will reduce the diastolic flow component.
75
Fig. 8. A freeze frame DS image of the left main coronary artery at its origin
with the aortic root. In these preliminary studies, the SV is trached
in the LMCA thus allowing only blood velocity spectra and not wall
reflections to be recorded throughout the cardiac motion. Note the
predominant diastolic flow velocity (R is the R wave of the EKG). It is
hoped that lesions of the LMCA can be detected and noninvasive
estimates of LMCA flow variables can be made.
ferent patient groups are given in Fig. 6. Results using the DS
with FFT spectral analysis obtained in 1980 are compared to
results using simple nonimaging CW audio spectral analysis
in 1979 and results using oculoplethysmography (OPG) in
1978. These results suggest that relatively inexpensive CW
methods with careful operator training and Doppler audio
spectral analysis can significantly increase the diagnostic ef-
ficacy of the noninvasive cerebrovascular lab with OPG on-
ly. More costly, yet more versatile, imaging DS methods can
further improve (but not significantly) the overall accuracy of
the noninvasive detection of lesions of >20% diameter
reduction in the extracranial vasculature. Importantly, the
percentage of negative angiographic procedures decreased
significantly with implementation of either CW or DS
methods. These results suggest that only patients with a
high probability of carotid disease are subjected to the risks
and costs of angiography.
Upon consideration of the good clinical results published
by others and results from our laboratory using the DS in
carotid disease, an important question arises [1]. If adequate
noninvasive diagnostic procedures which lessen the need
for angiography and provide unique qualitative physiologi-
cal information can be obtained with simple Doppler equip-
ment, why develop expensive DS systems which should not
and will not be competitive with standard arterial injection
or intravenous digital angiography [22] in the appropriate
workup of patients presenting symptoms of vascular insuffi-
ciency? In the experience of this laboratory, the answer to
this question is twofold. First, due to its unique imaging and
range-gating Doppler capabilities, a DS system with variable
transmit frequencies can be used noninvasively to detect
and quantitate anatomical lesions not only in the carotids
and femorals but also in other segments of the cardiovascu-
lar system which are not surrounded by bone or air. The
same DS unit can be used to examine intracardiac flow
regimes as well as systemic vascular lesions. As shown in
76
NONINVASIVE ECHO DOPPLER ULTRASOUND
FLOWMETRY
HZaHV
SUlsI*
mw1
-eA-
PU MONARY
I [T71s
RENAL
- iT I
UT.
____.
^Wor'&'ltl^lU
Fig. 9. A schematic representation of the various cardiovas cular applica- tle operator training and expertise are required for reliable
tions of a single, multifrequency DS. Not shown her(e are its addi- results. Standardization of the examination procedure and
tional intracardiac applications and its ability to measure cardiac the interpretation of results will further enhance the utility of
output. Note the variations of the blood velocit waveforms
y
depending on the input impedance to different v ascular beds.
These results suggest that image guided Doppler spectra and
calculated hemodynamic variables obtained from the DS can be us-
ed to detect anatomical lesions and to quant ify the
hemodynamic significance of these lesions througl iout the car-
diovascular system.
Fig. 7, application of the DS to the renal arterie .s has been deep-lying
as been 1) resolution, 2) fields of view, and 3) penetration, ultrasonic
reported [35]. Attempts to noninvasively di
ntect
flow images [30] are unlikely to provide the extensive anatomical
characteristics of the left main coronary artery usi ng tracking detail of either standard or
devices [21] are presently being undertake
laboratory (Fig. 8). Fig. 9 illustrates the various aenplictinons
of DS to obtain vessel images and calibrated blotopplications
signals (and subsequent flow rates) throughout tod
velocity
vascular system. Spectral analysis techniques to identify
arterial lesions can be applied to any of the ve.ssels. Gray
scale [18] and color-coded, real-time, multigaited, echo-
Doppler images [23] have also been develope u irom u1 J -
technology to demonstrate the immense amoun t of physio-
logical information and display formats that can b)e obtained
from pulsed Doppler ultrasound.
Second, it is the opinion of this laboratory that technologi-
cal innovations associated with the DS have allov ved reliable
noninvasive estimates of cardiac output [42] ar id regional
blood flow rates [32]. These measurements will become in-
creasingly important for the following reasons: The hemo-
dynamic severity and clinical consequences of irnpared gas
exchange due to an anatomical lesion depencI on many
physiological factors [64]. Unfortunately, these fiactors have
not been quantitatively related to the ischemic sy 'mptoms or
IEEE Transactions on Medical Imaging, Vol. MI-1, No. 1, July 1982
metabolic disorders [59]. In our view, a uniplane or biplane
anatomical description of an atherosclerotic lesion is impor-
tant but not sufficient information in the understanding and
quantitative evaluation of atherosclerosis. Hopefully, the
development of computer controlled videodensitometry
method will allow quantitative blood flow information to be
AN
obtained during angiography. Until then, more specified
anatomical and dynamic physiological information may be
obtained noninvasively (and without radiation exposure) in
the human from the DS.
Conclusions and Suggestions
Clearly, the simple continuous wave velocity detector
with or without sophisticated audio spectral analysis will re-
tain and increase its role as an important noninvasive tool in
the qualitative evaluation of cerebral and peripheral
vascular disease. No significant modifications of these
devices and methods which would not decrease their favor-
able cost-effectiveness seem warranted.
The diagnostic efficacy of CW and PD flow imaging devices
generally determined by the capability of Doppler spec-
are
tral analysis to reflect flow disorders secondary to anatomic
lesions. These relatively inexpensive imaging units are quite
specialized only to carotid examinations, but they appear to
adequately separate occlusive, less than 50% diameter re-
duction, and greater than 50% diameter reduction internal
carotid lesions. Compared to other modalities, relatively lit-
these modalities.
In the extracranial vascular system, duplex scanning
methods appear equally precise as angiographic interpreta-
tions in the quantification of atherosclerotic lesions. This
true
capability should extend to applications in other superficial
of
pure
and vessels. Due to technical constraints
With im- digital angiography.
proved transducer design, sample volume control [60], and
digital coding, adequate vessel images and Doppler spectra
can be recorded to provide the acquisition of unique blood
velocity and flow data. Consequently, a better clinical and
pathophysiological understanding of the true hemodynamic
severity of an atherosclerotic lesion can be obtained.
AL.A..
The authors would like to thank Drs. U. C. Luft, Jack Loep-
pky, Neal Halpern, Jonathan Sands, Robert jahnke, David
Sommerville, Robert Croke, Emmett Mathews, Pratap
Avasthi, David Hoekenga, and Richard Conn for their profes-
sional expertise. John Adams, Treva Miller, Pat Reilly, Arthur
Witt, Ruth Graham, Linda Bernhardt, and Isidora Miranda
provided excellent technical support. The work was sup-
ported by grants from the American Heart Association and
NHLBI grants 5-R01HL26025- 02 and 1-R01HL27095-01.
References
[1] R. H. Ackerman, A pespective on noninvasive diagnosis of carotid
disease. Neurology, vol. 29, pp. 615-22, 1979.
Greene et al.: Quantitative Evaluation of Atherosclerosis
[2] D. W. Baker, Pulsed ultrasonic Doppler blood flow sensing. IEEE SU - [29] R. W. Gill, Pulsed Doppler with B-mode imaging for quantitative blood
vol. 17, pp. 170-185, 1970.
[3] D. W. Baker, Applications of pulsed Doppler techniques. Radiol. Clin.
North Am., vol. 18(1), pp. 79-103, 1980.
[4] D. W. Baker and R. E. Daigle, Noninvasive ultrasonic flowmetry. In
Cardiovascular Flow Dynamics and Measurements, N. H. Hwang and
N. Norman, Eds. Baltimore, University Park Press, pp. 151-189, 1979.
[5] D. E. Baker, S. L. Johnson and D. E. Strandness, Prospects for quantita-
tion of transcutaneous pulsed Doppler techniques in cardiology and
peripheral vascular disease. In Cardiovascular Applications of Ultra-
sound; R. S. Reneman, Ed. Amsterdam, North Holland, vol. 108, p. 24,
1974.
[6] F. E. Barber, D. W. Baker, A. W. Nation, D. E. Strandness and J. M.
Reid, Ultrasonic duplex echo-Doppler scanner. IEEE Trans. Biomed.
Eng., vol. 21, p. 109, 1974.
[7] R. W. Barnes, Office Doppler techniques in vascular disease. I. of
Family Practice, vol. 13, pp. 711-720, 1981.
[8] R. W. Barnes, L. Nix and S. E. Rittgers, Audible interpretation of carotid
Doppler signals. Arch. Surg., vol. 116, pp. 1185-1189, 1981.
[9] R. W. Barnes, S. E. Rittgers and W. W. Putney, Real-time Doppler
spectrum analysis. Arch. Surg., vol. 117, pp. 52-57, 1982.
[10] W. S. Bartynski, P. Darbouze and P. Nemir, Jr., Significance of
ulcerated plaque in transient cerebral ischemia. Am. J. Surg., vol. 141,
pp. 353-357, 1981.
[11] S. M. Berry, J. A. O'Donnell and R. W. Hobson, Capabilities and limita-
tions of pulsed Doppler sonography in carotid imaging. J. Clin. Ultra-
sound, vol. 8, pp. 405-412, 1980.
[12] W. M. Blackshear, D. J. Phillips, P. M. Chikos, J. D. Harley, B. L. Thiele
and D. E. Strandness, Carotid artery velocity patterns in normal and
stenotic vessels. Stroke, vol. 11, pp. 67-71, 1980.
[13] W. M. Blackshear, Jr., D. J. Phillips, and D. E. Strandness, Jr., Pulsed
Doppler assessment of normal human femoral artery velocity pat-
terns. J. Surg. Res. vol. 27, pp. 73-83, 1979.
[14] W. M. Blackshear,Jr., D. J. Phillips, B. L. Thiele, et al., Detection of
carotid occlusive disease by ultrasonic imaging and pulsed Doppler
spectrum analysis. Surgery, vol. 86(5), pp. 698-706, 1979.
[15] K. C. Bodily, R. E. Zierler, M. R. Marinelli, B. L. Thiele, F. M. Greene, Jr.,
and D. E. Strandness, Jr., Flow disturbances following carotid en-
darterectomy. Surg. Cynecol. Obstet. vol. 151, pp. 77-80, 1980.
[16] G. E. Bone and R. W. Barnes, Limitation of the Doppler
cerebrovascular examination in hemispheric cerebral ischemia.
Surgery, vol. 79, p. 577, 1976.
[17] P. J. Breslau, G. Fell, D. J. Phillips, B. L. Thiele and D. E. Strandness, Jr.,
Evaluation of carotid bifurcation disease. Arch. Surg., vol. 117, pp.
58-60, 1982.
[18] D. J. Burch and J. C. Taenger, Real-time quantitative blood flow
measurements using Doppler ultrasound. Proceedings 26th AIUM, San
Francisco, CA, p. 17, 1981.
[19] A. J. Comerota, J. J. Cranley and S. E. Cook, Real-time B-mode carotid
imaging in diagnosis of cerebrovascular disease. Surgery, vol. 89(6),
pp. 718-729, 1981.
[20] P. L. Cooperberg, W. D. Robertson, P. Fry and V. Sweeney, High
resolution real time ultrasound of the carotid bifurcation. J. Clin. Ultra-
sound, vol. 7, pp. 13-17, 1979.
[21] J. C. Davis, K. L. Richards and E. R. Greene, A sample volume tracking
unit for pulsed Doppler echocardiography. IEEE Trans. Biomed. Eng.,
vol. 26, pp. 285-288, 1979.
[22] M. Ducos de Lahitte, J. P. Marc-Vergnes, A. Rascol, B. Guiraud and C.
Manelfe, Intravenous angiography of the extracranial cerebral arteries.
Radiology, vol. 137, pp. 705-711, 1980.
[23] M. K. Eyer, M. A. Brandestini, D. J. Phillips and D. W. Baker, Color
digital echo/Doppler image presentation. Ultrasound Med. Biol., vol.
7(1), pp. 21-31, 1981.
[24] H. Feigenbaum, Echocardiography, 3rd Ed. Philadelphia, Lea and
Febiger, 1981.
[25] G. Fell, D. J. Phillips, P. M. Chikos, J. D. Harley, B. L. Thiele and D. E.
Strandness, Jr., Ultrasonic duplex scanning for disease of the carotid
artery. Circulation, vol. 64(6), pp. 1191-1195, 1981.
[26] S. W. Flax, J. G. Webster and S. J. Updike, Statistical evaluation of the
Doppler ultrasonic blood flowmeter. Biom. Sci. Instr., vol. 7, pp.
201-222, 1970.
[27] D. L. Franklin, W. A. Schlegel and R. F. Rushmer, Blood flow measured
by Doppler frequency shift of backscattered ultrasound. Science, vol.
132, pp. 564-565, 1961.
[28] D. 0. Gidden, R. E. Mabon and R. A. Cassanova, Measurement of
disordered flows distal to subtotal vascular stenoses in the thoracic
aortas of dogs. Circ. Res., vol. 39, pp. 112-119, 1976.
77
flow measurement. Ultrasound Med. Biol., vol. 5(3), pp. 223-235,
1979.
[30] A. Goldstein, Range ambiguities in real-time ultrasound.J. Clin. Ultra-
sound, vol. 9, pp. 83-90, 1981.
[31] R. G. Gosling, Extraction of physiological information from spectra
analyzed Doppler-shifted continuous wave ultrasound signal obtained
noninvasively from the arterial system. IEEE Medical Electronics
Monograph 21. Hill, Walson, Eds. Peter Peregrinius, 1976, pp. 73-125.
[32] E. R. Green, Noninvasive measurement of blood flow using pulsed
Doppler ultrasound. In Oxygen Transport to Human Tissues; J. A.
Leoppky, Ed. New York, Elsevier, 1981.
[33[ E. R. Greene and M. B. Histand, Ultrasonic assessment of simulated
atherosclerosis: in vitro and in vivo comparisons. I. Biomech. Eng., vol.
101, pp. 73-81, 1979.
[34] E. R. Greene, D. E. Hoekenga, K. L. Richards and J. G. Davis, Pulsed
Doppler echocardiographic audio spectrum analysis: time interval
histogram versus multifilters spectrogram and fast Fourier transform.
Biomed. Sci. lnstr., vol. 16, pp. 134-144, 1980.
[35] E. R. Greene, M. D. Venters, P. S. Avasthi, R. L. Conn and R. W. Jahnke,
Noninvasive characterization of renal artery blood flow. Kidney Int.,
vol. 20, pp. 523-529, 1981.
[36] C. J. Hartley and D. E. Strandness, The effects of atherosclerosis on
transmission of ultrasound. I. Surg. Res., vol. 9, pp. 575-582, 1969.
[37] R. W. Hobson, S. M. Berry, A. S. Katocs, Jr., J. A. O'Donnell, Z. Jamil
and J. P. Savitsky, Comparison of pulsed Doppler and real-time
B-mode echo arteriography for noninvasive imaging of the ex-
tracranial carotid arteries. Surgery, vol. 87(3), pp. 286-293, 1980.
[38] D. E. Hokanson, D. J. Mozersky, D. S. Sumner, F. D. McLeod and D. E.
Strandness, Ultrasonic arteriography: a noninvasive method for
arterial visualization. Radiology, vol. 102, pp. 435-436, 1972.
[39] J. E. Jorgensen, D. Campau and D. W. Baker, Physical characteristics
and mathematical modelling of the pulsed ultrasonic flowmeter. Med.
Biol. Eng., vol. 2, pp. 404-421, 1973.
[40] B. A. Keagy, W. F. Pharr, D. Thomas and D. E. Bowes, A quantitation
method for the evaluation of spectral analysis patterns in carotid
artery stenosis. Ultrasound Med. Biol: In press, 1982.
[41] R. R. Lewis, M. G. Beasley, D. E. Hyams and R. G. Gosling, Imaging the
carotid bifurcation using continuous-wave Doppler-shift ultrasound
and spectral analysis. Stroke, vol. 9(5), pp. 465-471, 1978.
[42] J. A. Loeppky, E. R. Greene, D. E. Koekenga, A. Caprihan and U. C.
Luft, Beat-by-beat stroke volume assessment by pulsed Doppler in
upright and supine exercise. 1. AppI. Physiol., vol. 50, pp. 1173-1182,
1981.
[43] M. H. Lunt, Accuracy and limitations of the ultrasonic Doppler
velocimeter and zero-crossing detector. Ultrasound Med. Biol., vol. 2,
pp. 1-10, 1975.
[44] M. R. Marinelli, K. W. Beach, M. J. Glass, J. F. Primozich and D. E.
Strandness, Jr., Noninvasive testing vs clinical evaluation of arterial
disease. IAMA, vol. 241, pp. 2031-2034, 1979.
[45] D. A. McDonald, Blood Flow in Arteries, 2nd Ed. Baltimore, William
and Wilkins, 1974.
[46] A. Novicki and J. M. Reid, An infinite gate pulse Doppler. Ultrasound
Med. Biol., vol. 7(1), pp. 41-50, 1981.
[47] 0. H. O'Leary, M. E. Clouse, A. V. Persson and S. A. Edwards, Nonin-
vasive testing for carotid artery stenosis. II Clinical application of ac-
curacy assessments. AIR, vol. 138, pp. 109-111, 1982.
[48] M. S. Pessin, G. W. Duncan, J. P. Mohr and D. C. Poskanzer, Clinical
and angiographic features of carotid transient ischemic attacks. N.
Engl. J Med., vol. 296(7), pp. 358-362, 1977.
[49] D. J. Phillips, J. E. Powers, W. M. Eyer, et al., Detection of peripheral
vascular disease using the duplex scanner Ili. Ultrasound Med. Biol.,
vol. 5, pp. 205-218, 1980.
[50] J. L. Prendes, W. M. McKinney, F. S. Buonanno and A. N. Jones,
Anatomic variations of the carotid bifurcation affecting Doppler scan
interpretation. I. Clin. Ultrasound, vol. 8, pp. 147-150, 1980.
[51] J. M. Reid and M. P. Spencer, Ultrasonic Doppler technique for imag-
ing blood vessels. Science, vol. 176, pp. 1235-1236, 1972.
[52] R. S. Reneman and M. P. Spencer, Difficulties in processing of an
analogue Doppler flow signal: with special reference to zero-crossing
meter and quantification. In Cardiovascular Applications of Ultra-
sound R. S. Reneman, Ed. Amsterdam, North Holland, 1974.
[53] R. S. Reneman and M. P. Spencer, Local Doppler audio spectra in nor-
mal and stenosed carotid arteries in man. Ultrasound Med. Biol., vol.
5, pp. 1-11, 1979.
[54] S. E. Rittgers, W. M. Putney and R. W. Barnes, Real-time spectrum
analysis and display of directional Doppler ultrasound blood velocity
78
signals. IEEE Trans. Biomed. Eng., vol. 27, pp. 723-728, 1980.
[55] R. B. Rutherford, W. R. Hiatt and E. W. Kreutzer, The use of velocity
waveform analysis in the diagnosis of carotid artery occlusive disease.
Surgery, vol. 82, pp. 695-702, 1077.
(56] R. D. Shoumaker and S. Bloch, Cerebrovascular evaluation: assess-
ment of Doppler scanning of carotid arteries, ophthalmic Doppler
flow and cervical bruits. Stroke, vol. 9(6), pp. 563-566, 1977.
[57] M. P. Spencer, J. M. Reid, Quantitation of carotid stenosis with
continuous-wave (C-W) Doppler ultrasound. Stroke, vol. 10(3), pp.
326-330, 1979., vascular disease. Prog. Cardiovasc. Dis., vol. 20(6), pp.
403-422. 1978.
[58] D. E. Strandness, Jr., The use of ultrasound in the evaluation of
peripheral vascular disease. Prog. Cardiovasc. Dis., vol. 20(6), pp.
402-422, 1978.
[59] T. M. Sundt, F. W. Sharbrough, D. G. Piepgras, T. P. Kearns, J. M.
Messick and W. M. O'Fallen, Correlation of cerebral blood flow and
electrocephalographic changes during carotid endarterectomy. Mayo
IEEE Transactions on Medical Imaging, Vol. MI-1, No. 1, July 1982
Clinic Proceedings, vol. 56, pp. 533-543, 1981.
[601 A. R. Walker, D. J. Phillips and J. E. Powers, Evaluating Doppler
devices using a moving string test target. J. Clin. Ultrasound, vol. 10,
pp. 25-30, 1982.
[61] R. G. Weaver, Jr., C. Howard, W. M. McKinney, M. R. Ball, A. M. Jones
and J. F. Toole, Comparison of Doppler ultrasonography with
arteriography of the carotid artery bifurcation. Stroke, vol. 11(4), pp.
402-404, 1980.
[62] P. N. T. Wells, Biomedical Ultrasonics. New York, Academic Press,
1977.
[63] D. White and G. Curry, Color-coded differential Doppler ultrasonic
scanning system for the carotid bifurcation: Results in 486 bifurcation
angiographically confirmed. In Recent Advances in Ultrasound
Diagnosis; A. Kurjak, Ed. Amsterdam, Excerpta Medica, 1978, pp.
239-249.
[64] D. F. Young, Fluid mechanics of arterial stenoses. Journal of
Biomechanical Engineering, vol. 101, pp. 157-175, 1979.