Recently there has been a dramatic increase in the. Performance Tests of Doppler Ultrasound Equipment With a Tissue and Blood Mimicking Phantom

Transcription

1 Performance Tests of Doppler Ultrasound Equipment With a Tissue and Blood Mimicking Phantom Evan J. Boote, MS, James A. Zagzebski, PhD A tissue- and blood-mimicking phantom was assembled for assessing the performance of ultrasound Doppler equipment. The phantom is in the shape of a rectangular parallelepiped with a 10 x 20 em scanning window and a depth of 16 em. Components of the phantom include a tissue-mimicking material, 7.9 mm diameter simulated vessels, a fluid with similar back scatter as whole blood, and a peristaltic pumping system producing peak scatterer velocities greater than 1 mjsec. Performance tests done with -the phantom are outlined. These include assessments of the maximum depth of penetration and of the directional discrimination capabilities of the instrument, determinations of the accuracy of displayed flow velocities, and accuracy assessments of the displayed position of the Doppler sample volume. KEY WORDS: Doppler; ultrasound; quality assur ance; performance; phantom. (/Ultrasound Med 7:137, 1988) Recently there has been a dramatic increase in the clinical use of ultrasound instruments that use the Doppler effect to display flow information.t-3 This is due to both more widespread use of stand-alone Doppler systems and the introduction of duplex instruments that display a 2-dimensional pulseecho image with a Doppler signal frequency spectrum. With such instruments it is possible to record and display blood flow velocities from a selected, localized region in the body. 4 As with any complex system, it is desirable to have methods for objectively assessing the performance of ultrasound Doppler instruments. A variety of test methods have been proposed, many including the use of oscillating targets 5.6 or flowing liquids 7.8 to evaluate an Received January 1, 1987, from the University of Wisconsin Madi son, Department of Medical Physics, Madison, Wisconsin. Revised manuscript accepted for publication July 24, Address correspondence and reprint requests to Dr. Boote: Univer sity of Wisconsin-Madison, Dept. of Medical Physics, 1530 Medical Sciences Center, Madison, WI instrument's temporal and spatial response to moving reflectors. Such test methods are useful for characterizing some factors related to the performance of an instrument, eg, the velocity readout accuracy and the spatial resolution. However, they do not always provide information that is easily interpreted in terms of clinical results because echo signal levels from some test objects are much greater than signals from targets in the body. Consequently, performance variables that may be closely related to the detected signal level, such as the sensitivity of the system and the directional discrimination capabilities of directional Doppler instruments, are not easily assessed with such devices. The purpose of this report is to describe a phantom and the methods for evaluating the performance of Doppler ultrasound equipment. Emphasis is given to duplex instruments, although many of the tests also are applicable to 2-dimensional color flow imaging devices and to stand-alone Doppler devices. Quantitative tests that have been developed for use with this phantom and may be carried out by clinical personnel are described. Preliminary results from several instruments also are presented by the American Institute of Ultrasound in Medicine J Ultrasound Med 7: , ij'88j$3.50

2 138 DOPPLER ULTRASOUND EQUIPMENT l Ultrasound Med 7: , 1988 DESCRIPTION OF THE PHANTOM Design goals of the phantom were to provide reproducible and well-characterized test conditions that closely simulate the clinical environment and challenge the capabilities of Doppler instruments. Important components of the Doppler phantom are as follows: 1) a stable blood-mimicking fluid that can be imaged at different depths in a tissue-like medium; the fluid has ultrasonic scattering properties that are the same as human blood at the frequencies of concern. The fluid may be pumped at various speeds without introducing unwanted air bubbles. And 2) a tissue-mimicking path between the transducer and the fluid which adequately represents regions that are studied with the Doppler instrument; the path distance from the transducer to the vessel is variable. The specific phantom configuration used in the present study is diagramed in Figure 1. The phantom is in the shape of a rectangular parallelepiped with a 10 x 20 em scanning window at the top and has a depth of 16 em. It contains tissue-mimicking material made of a water-based gel with a uniform distribution of graphite powder to control the attenuation. 9 The attenuation and speed of sound of the tissue-mimicking material are similar to those of liver tissue. The attenuation coeffi cient is 0.5 dbfcm at 1 MHz and is proportional to the ultrasonic frequency. The speed of sound is 1540 mfsec. A latex rubber vessel with an inner diameter of 7.9 mm and a wall thickness of 1.6 mm runs through the middle Figure 1 The Doppler ultrasound phantom used in this work. Latex vessels (one shown) are embedded in a box of tissue-mimicking gel with an attenuation coefficient of 0.5 db/em at 1 MHz and is proportional to frequency. The speed of sound in the material is 1540 mfsec. The acoustical scanning surface is at the top of the phantom, presenting the latex vessel at the 45 angle to the transducer. A blood-mim icking fluid circulates through the dosed ]oop. pumped by a variable speed peristaltic pump. Roservalt of the phantom at a 45 angle with respect to the scanning window. A second vessel (not shown) runs horizontally through the phantom at a depth of 6 em. Circulating through these vessels is a blood-mimicking fluid. The fluid is a mixture of degassed water and glycerol in correct proportions to give a specific gravity of g/cm 3 This specific gravity was chosen toreduce sedimentation of the third component of the blood-mimicking fluid 1. polystyrene microspheres [Duke Scientific, Inc., Palo Alto, CA, Polystyrene DVB 30,um(nominal), catalog no. 2428]. The microspheres provide scattering from the fluid; their size distribution and concentration were selected (see below) to provide a backscatter level equivalent to that of actual blood in the 2.5 to 6 MHz frequency range. The speed of sound of this fluid is 1546 mfsec. The attenuation coefficient is 0.1 dbfcm at 1 MHz. The concentration of microspheres to yield the same backscatter levels as blood was found empirically using the apparatus shown in Figure 2. Outdated whole human blood having a hematocrit of 43 was circulated through a pancake-shaped reservoir having acoustic windows of Saran (Dow Chemical Corp., Midland, MI). Caution was exercised to assure that there was no clotting formation in the blood, producing an artificially high scatter level. The reservoir was placed in the beam of a 3.5 MHz transducer that was driven with a broadband pulse using a commercially available pulser receiver (Panametrics Model 5052UA, Waltham, MA). The same transducer was used to detect scattered echo signals from fluid within the reservoir. Echo voltage signals originating from the reservoir were selected by time gating; these signals were then applied to a spectrum analyzer (Tektronix Model 7U2, Beaverton, OR). The relative backscatter level of flowing blood was deter- Figure 2 Experimental apparatus used to determine relative backscaher levels of whole human b1ood and blood-mimick ing fluid. Arrangement of the pancake reservoir was identical in both cases. Water bath Scannln& lrindow Tluuc mlmlc:ldai plwltom V.n.bleapecd pcrtmltlc: pump

3 J Ultrasound Med 7: , 1988 BOOTE AND ZAGZEBSKI 139 mined from the mean spectral response for 30 independent echo signal waveforms. The blood-mimicking fluid described earlier was then circulated through the reservoir. The concentration of microspheres for the blood-mimicking material was found by adding microspheres until the mean backscatter level was within 3 db of that for whole blood at all frequencies in the bandpass of the pulsed transducer (Fig. 3). For 30 J.l diameter polystyrene microspheres, a concentration of 17 X 10 4 em -l yields similar scattering as that of blood. Because the microspheres are Rayleigh scatterers (as well as red blood cells) they exhibit the same frequency response of backscatter as red blood cells. The concentration of the microspheres in the tissue-mimicking blood is much lower than that of red blood cells in whole blood, mainly because of the use of 30 J.l diameter particles and the fact that the cross section for Rayleigh scattering is proportional to the sixth power of the scatterer diameter. The pumping system is a closed loop of two different diameter vessels. The smaller vessels reside in the phantom and, as stated previousl~ are 7.9 mm in diameter. A larger vessel is placed within a peristaltic pump (Little Giant Pump Co., Model LG-300, catalog no , Oklahoma Cit~ OK) driven by a variable speed motor [Cole-Parmer Instrument Co., Masterflex, catalog no. J , Chicago, IL (details of the pumping system are given because of this writing, the authors are not aware of any equivalent commercially available system)]. This vessel has an inside diameter of 11.1 mm. The two Figure 3 Relative backscatter level: blood-mimicking Ouid and whole human blood, plotted as a function of ultrasonic frequency. The backscatter levels are compared by subtracting the mean scattered signal amplitude from blood at a given frequency from the corresponding signal amplitude from blood-mimicking fluid at the same frequency /II -- I - I --- i I ~ Frequene1 (MJfz) vessels are connected by a section that is tapered hom the 11.1 mm diameter end to the 7.9 mm diameter end, avoiding turbulence at that junction. A reservoir approximately 500 ml in volume collects the fluid after it has passed through the phantom. The use of these components allows fluid velocities approaching 150 cmfsec without introducing air bubbles into the fluid. Any induced air bubbles would defeat the purpose of having a blood-mimicking fluid by causing an extremely high backscatter level with reference to blood. This system produces basically a rectangular flow velocity waveform with a 50% duty cycle. More specialized flow velocity waveforms might be obtained by using specifically designed cams along with a peristaltic pump 10 ; however, we have not introduced these into the present phantom. PERFORMANCE TESTS OF DOPPLER INSTRUMENTS This phantom has been used for evaluation and intercomparison of duplex Doppler instruments before equipment purchase. Performance factors that were considered include the sensitivity and noise level of the Doppler instrument, directional discrimination capabilities, accuracy of the displayed velocities, and the accuracy of the displayed position of the Doppler sample volume. Test procedures using the phantom are described below. Maximum Penetration For Doppler signals within the frequency range of an instrument, the maximum sensitivity is a measure of the weakest Doppler shifted echo signals that the instru ment can detect and display at a satisfactory level above the electronic noise. Clinically, maximum sensitivity is most closely related to the ability to detect Doppler signals from small vessels located in an attenuating medium, large distances hom the transducer assembly. The extent to which this can be done is measured by a maximum penetration test into the Doppler flow phantom. Typically, the maximum penetration test is done with the output power turned to its maximum level and the Doppler receiver gain increased to a level without exces sive electronic noise on the display. Because the vessel is placed at a 45 angle with respect to the scanning win dow of the phantom, lateral movement of the transducer assembly allows one to increase the transducer assembly-to-vessel distance. Doppler signals from the region interrogated get progressively weaker at greater depths. Figure 4 shows Doppler spectral displays for varying tissue path lengths to the vessel lumen. It can be seen from the flow waveform that the Doppler signal strength diminishes to a point where the displayed flow signal

4 140 DOPPLER ULTRASOUND EQUIPMENT J Ultrasound Med 7: , 1988 Figure 4 Doppler spectral displays Cor varying tissue path lengths with a S MHz probe. A, shows the signal Cor a tissue path length of S em. 8, is for a path length of 6 em. Note how the ftow velocity display changes Cor longer tissue path lengths and thus higher attenuation. The reduction of the brightness of this display indicates a lower Doppler signal level. The maximum penetration for these test conditions and this system was 7.5 em. A B cannot be discriminated from the system electronic noise. The audible Doppler signal also diminishes until it too is indistinguishable from the noise. The maximum penetration for these test conditions and this system was approximately 7.5 em. The maximum range at which the Doppler signal is detected with the audible Doppler signals and the Doppler spectral display is noted. This is called the maximum penetration for that transducer assembly. Figure 5 presents results of this test for several different model sector scanners. Most had different frequency transducer assemblies available, allowing differences in penetration for different frequencies to be illustrated. There also appears to be some variation in penetration for transducers of the same frequency on instruments from different manufacturers. For example, unit Doperating at 25 MHz penetrates 15.5 em whereas unite penetrates to 14 em for the same frequency. These variations may be due to differences in maximum output power, different pulse durations, and overall gain and noise properties of different instruments. Units B and C show variations in penetration when different power settings

5 J Ultrasound Med 7: , 1988 BOOTE AND ZAGZEBSKI 141 Figure 4 (continued) Doppler spectral displays for varying tissue path lengths with a 5 MHz probe. C, shows the signal for a tissue path length of 7 em; Q is for a path length of 8 em. c D are observed at an operating frequency of 5 MHz. Unfortunately, the actual acoustic intensities produced by these instruments were not available at the time of this study to allow comparison between penetration and acoustic intensity. Channel Isolation or Directional Discrimination Directional Doppler instruments should allow the interpreter to determine whether flow is moving towards or away from the transducer. Usually, flow direction is determined in the instrument by forming the Doppler signal into two separate electrical channels that are in qua- drature, or 90 (0 out of phase. The instrument's spectral display presents flow in either a positive or negative direction, depending on the relative phase of the signals in these two channels. Inadequate discrimination between signals in these channels may result in mirroring of the spectral display about the baseline (Fig. 6), presenting the appearance of bidirectional or turbulent flow even though the flow is unidirectional. This effect is more evident when strong Doppler signals are present. To assess directional discrimination, the Doppler sample volume cursor is positioned within the vessel using a transducer beam orientation that will assure

6 142 DOPPLER ULTRASOUND EQUIPMENT J Ultrasound Med 7: , 1988! II 1... i 0 t! I 8 ~ , A Doppler Ult.JUOwsd unlta Figure 5 Summary of results of maximum depth of penetra tion for five different Doppler ultrasound scanners. Results for four frequencies are shown; however, not all the units had each frequency available. Lower frequency ultrasound pulses result in greater capability to penetrate and return a Doppler signal. Note how the low power setting for 5 MHz is not capable of penetrating as deeply as the high power setting. 1m, 5 MHz low power; ml, 5 MHz high power; El, 3.5 MHz,;, 2.5 MHz. detection of signals from reflectors moving in one direction only. The flow velocity should be low enough that aliasing does not occur (see below). The display is then examined to determine whether any noticeable leakage of signal occurs, making it appear that flow is bidirec- tional. By placing the transducer assembly and sample volume to detect flow at different depths, discrimination for different strength Doppler signals can be tested. Accuracy of Doppler Sample Volume Cursor for Duplex Instruments Another test is concerned with the accuracy to which the position of the sample volume cursor (found on most duplex Doppler instruments) is represented on the B mode image. The test consists of positioning the transducer assembly so that the vessel is imaged and recording a Doppler signal spectrum as the sample cursor is slowly moved across the vessel. The strongest Doppler signal is expected when the cursor appears at the center of the vessel (Fig. 7A). On units were the B-mode display of the sample volume position is inaccurate, the strongest signal may be present when the cursor appears off center or, at worst, on the edge or outside the vessel (Fig. 7B). The importance of this test is obvious when one considers the difficulties associated with exact placement of the Doppler sample volume in some small vessels. In order to avoid erroneous results, it is important to verify that the speed of sound assumed in the calibration of the ultrasound instrument is the same as that in the phantom. This is especially true for duplex transducer assemblies that use separate transducers for B-mode imaging and Doppler. For example, in imaging applications where the tissue path is mainly muscle 1 higher speeds of sound than that of the present phantom may be assumed in the position computation circuitry of the instrument. Figure 6 This figure is an example of a directional Doppler instrument that has poor channel isola tion. The mirroring of the Doppler spectral display about the baseline when flow is laminar and unidirectional is an indication of poor directional discrimination.

7 J Ultrasound Med 7: , 1988 BOOTE AND ZAGZEBSKI 143 Figure 7 Tests of the displayed accuracy of the Doppler sample volume cursor. This figure demonstrates a unit which has proper positioning (on the B-mode image) of the Doppler sample volume cursor (A), and a unit which errors in the displayed position of the Doppler sample volume cursor (8). A, the strongest Doppler signal is seen when the cursor is positioned in the center of the vessel. 8, shows a unit in which the strongest Doppler returns are not found when the cursor is centered on the vessel, but when the cursor is positioned (on the image) off the edge of the vessel. A B

8 144 DOPPLER ULTRASOUND EQUIPMENT J Ultrasound Med 7: , 1988 Flow Velocity Readout An important application of Bow phantoms is to determine the accuracy of the velocity (or Bow rate) readout ~f a Doppler instrument. This requires the phantom s pump speed control to be calibrated so that absolute rebectorjscatterer velocities or the absolute flow rates are known. In some clinical situations, the volume flow rate (ie, mlfsec) is of great interest. u When this is the c~e, calibration can be done easily if the phantom prov1des continuous flow. Simple collection in a volume flask or use of an in-line flow meter is sufficient for measuring flow rates as long as the Bow resistance conditions are the same as for when the phantom is in use. In the clinical applications in our department, the flow velocity (scatterer velocity in cmfsec) is of great interest. The flow velocity could be computed from the volume Bow rate if the velocity profile across the vessel is known. We, however, chose to measure the peak velocity directly. In our application of the?hantom~ the actual peak flow velocity vs pump speed as determmed by using a continuous wave peripheral vascular Doppler instrument. Testing of the duplex instrument of interest is then accomplished by positioning the sample volume cursor at a constant depth within the phantom and the flow waveforms represented by the instrument are recorded as a function of the pump speed. Figure 8 is an example of results from this test performed on a clinical duplex Doppler unit. Typically, the peak flow velocity measured Figure 8 Results of a flow velocity readout test. The flow measured by a continuous wave Doppler device as compared to that measured by one of the instruments under test. 120 loo f J 1'11 80 li 9 8 I I.._ meaunod ftlocllj' [cm/.el ColltlnUOUII- Dapplu with a duplex pulsed wave Doppler device was within 15% of the measured value using the continuous wave device. It is useful to perform periodic checks on Doppler instruments to assure accurate flow velocity readout (:tlo to 15%) before clinical use. Other Tests and Qualitative Impressions The phantom is convenient for teaching principles of Doppler instruments and for equipment demonstrations. For example, one of the factors that can be demonstrated is how a particular instrument displays Doppler signals in the presence of aliasing. Aliasing is the generation of artifactual, low frequency signals when the Doppler signal frequency exceeds the Nyquist limit, which is taken as one-half the pulse repetition frequency (PRF) of the Doppler ins~ment: The PR~ ca~ be adjusted on most instruments, e1ther directly or mdjrectly by varying the Doppler sample depth or the velocity range on the Doppler spectral display. Aliasing is often associated with a wrapping around of the spectral display (Fig. 9Q. However, some instruments exhibit a wrapping of the display even though the Nyquist limit has not been reached. Figure 9A shows a spectral display with flow activated in the vessel. The PRF is high enough so that the Doppler signals are adequately sampled. Figure 9B is an example of spectral display wrap-around, even though aliasing is not present. Flow conditions and the instruments PRF are identical to those in the previous case; however, the spectral baseline has been changed s~ the actu?l vel~city exceeds the maximum that can be displayed m a gwen direction. The conditions in Figure 9C are identical to Figure 9B except the spectral range has been decreased, which was accompanied by a decrease in the PRF. High frequency Doppler signals have been undersampled and appear as lower frequencies. It was fairly easy to detect this by listening to the Doppler audio signal because a distinctive change in this signal was noted when the spectral range (and hence the PRF) was decreased. We have observed that the spectral display changes are often more subtle than the effect on the audio signal when aliasing occurs. The controlled environment of a phantom is most convenient for this demonstration. Other aesthetic features also may be demonstrated with the phantom. These include ease of instrument operation, scanning menus available, and archiving scans. The phantom allows the evaluation of these items without the encumbrance of having to scan a patient. Additionally, acoustic exposure to patients and normal subjects is reduced or avoided, both in the process of evaluating the instrument and during the orientation/ teaching periods.

9 J Ultrasound Med 7: , 1988 BOOTE AND ZAGZEBSKI 145 Figure 9 A demonstration of spectral wrapping and aliasing. A, the Oow is seen on the Doppler spectral display with the baseline at the bottom of the vertical axis. The flow peaks at approximately 1 mjsec. B, the baseline has been moved up the vertical axis and the peak displayed positive flow is about 0.5 mjsec. The pulse repetition frequency has not changed and thus the sampling is sufficient to receive the entire spectrum of Doppler shifted frequencies. However, the display wrapsaround because the display positive flow portion of the display is exceeded. A 8 DISCUSSION A possible weakness of the protocol just described is the degree of subjectivity when establishing performance criteria for an instrument. However, for routine performance testing in a quality-controlled program, this subjectivity is of less significance, providing the same individual carrys out the tests from time to time. In addition, objective criteria may be incorporated if desirable. For example, in the maximum penetration and the directional discrimination tests, a root mean square voltmeter may be used to measure the output Doppler signal-to-noise ratio or the relative forward and reverse signals. Alternatively, some instruments provide velocity calculation algorithms as well as algorithms for computing characteristics of the flow, such as the pulsatility index. When the Doppler signal strength gets too weak to provide accurate indications of these parameters (eg, ±10% errors), this may also be used to locate the maximum penetration distance. Before using any phantom for measuring the accuracy of flow readout in a Doppler instrument, it is important

10 146 DOPPLER ULTRASOUND EQUIPMENT J Ultrasound Med 7: , 1988 Figure 9 (continued) C, the spectral range has been decreased. or, the pulse repetition frequency has been decreased so that the sampling is below the Nyquist frequency and the high frequency Doppler shifts are manifested as lower frequencies. c to verify that the phantom's flow rate and resultant peak scatterer velocity are correlated with the pumping speed of the phantom. We chose to calibrate the peak scatterer velocity in the phantom vs pump speed using a continuous wave Doppler instrument. Alternatively, volume collecb!d with a vessel over time or use of a string target test object could be used for this mesurement of Doppler instrument performance. There are several Doppler phantoms commercially available at this time; many employ centripetal pumps to circulate fluid through the phantom vessels. Such pumps have a tendency to introduce cavitation air bubbles into the fluid environment, especially at high pump speeds. The Doppler phantom presented here has a peristaltic pump that is able to pump high volume flows without the induction of air bubbles into the fluid. In combination with a fluid that has a scattering level similar to that of blood, the peristaltic pump allows tests of Doppler instrumentation under conditions that closely simulate clinical use of Doppler ultrasound. Other arrangements of vessel geometries and flow patterns may be useful, involving smaller vessels or clinically similar velocity flow patterns. However, the present phantom represents a simple configuration that allows testing of meaningful aspects of Doppler instrument performance. CONCLUSIONS We have developed a Doppler ultrasound phantom that employs a blood-mimicking fluid circulating through vessels in tissue-mimicking material Tests of Doppler instrument performance were described and prelimi- nary results of those tests presented. The phantom is also convenient for teaching principles of Doppler instruments and equipment demonstrations. ACKNOWLEDGMENTS The authors wish to acknowledge William Davros' and Michael lnsana's assistance in the development of synthetic blood and Gary Frank's work on mechanical aspects of the Doppler phantom system. The project was supported in part by a contract with Radiation Measurements, Inc. and by National Institutes of Health grant R01CA REFERENCES 1 Hatel, Uv, Angelson, Bjorn: Doppler Ultrasound in Cardiology. Physical Principles and Clinical Applications, Philadelphia, PA, Lee & Feiburger, Zweibel WJ, (ed): Introduction to Vascular Ultrasonography, 2nd edition. Grune & Stratton, Orlando, FL, Taylor KJW, Burns PN, Woodcock JP, et al: Blood flow in deep abdominal.and pelvic vessels: ultrasonic pulsed Doppler analysis Radiology 154:487, Gill RW: Measurement of blood flow by ultrasound: accuracy and sources of error. Ultrasound Med Bio ll:625, Walker AR, Phillips OJ, Powers JE: Evaluating Doppler devices using a string target. J Clin Ultrasound 10:25, Hoeks APG, Ruissen CJ, Hick P, et al: Methods to evaluate the sample volume of pulsed Doppler systems. Ultrasound Med Bio 10:427, 1984

11 1 Ultrasound Med 7: , 1988 BOOTE AND ZAGZEBSKI Reid JM: Methods of measuring the performance of continuous-wave Doppler diagnostic equipment. Draft IEC standard, Subcommitte 29D, Working group 10, McDicken WN: A versatile test-object for the calibration of ultrasonic Doppler flow instruments. Ultrasound Med Bio 12:245, Madsen EL, Zagzebski JA, Banjavic RA, et al: TISsue mimicking materials for ultrasound phantoms. Med Phys 5:391, McCarty K, Locke OJ: Test objects for the assessment of the performance of Doppler shift flowmeters. In Evans, JA (ed): Physics in Medical Ultrasound. London, Institute of Physical Sciences in Medicine, 1986, pp U. Gill RW: Pulsed Doppler with 8 -mode imaging for quantitative blood flow measurement. Ultrasound Med Bio 5:223, 1979