Myocardial regional blood flow: Quantitative measurement by computer analysis of contrast enhanced echocardiographic images

Ultrasound in Med. & BioL Vol. 19, No. 8, pp. 619-633, 1993

0301-5629/93 $6.00 + .00 © 1993 Pergamon Press Ltd.

Printed in the USA

OOriginal Contribution i

MYOCARDIAL REGIONAL BLOOD FLOW: QUANTITATIVE MEASUREMENT BY COMPUTER ANALYSIS OF CONTRAST ENHANCED ECHOCARDIOGRAPHIC IMAGES V I C T O R MoR-AVI,* D A N I E L DAVID,* S O L A N G E A K S E L R O D , t YVES B I T T O N ~ a n d ITZHAK CHOSHNIAK* *Medical Physics Laboratory, School of Physics, Tel Aviv University, tDepartment of Cardiology, Meir General Hospital, Kfar Saba, and *Department of Zoology, Faculty of Life Sciences, Tel Aviv University, Israel

(Received 12 October 1992; in final form 17 May 1993) Abstrnct--Quantitation of regional myocardial blood flow constitutes the missing link between the anatomy of coronary obstruction and its physiological effect on regional oxygensupply. Microscopic air bubbles, introduced into the coronary circulation, were shown to produce a transitory enhancement of the myocardial tissue contrast, easily detectable with standard ultrasonic imaging equipment. This study presents a new approach linking the tissue blood flow with the time-dependent changes in the intensity of the ultrasonic reflections produced by the microbubbles. The tissue blood flow is evaluated using the well-known indicator dilution relation, according to which flow equals the ratio between the intravascular fraction of the tissue sample volume and the mean transit time of the contrast agent. We derive these two parameters from the time curves representing the contrast induced variations in the mean videointensity measured in two regions of interest, a reference region in the left ventricular cavity and the region of interest within the myocardial tissue. The intravascular volume fraction is computed as the ratio of the total power of the above two intensity curves, as each of these is assumed to be proportional to the total amount of tracer traversing the corresponding region of interest. The mean transit time is computed using combined time- and frequency-domain processing, involving Fourier deconvolution of the response function of the myocardial tissue sample. This approach was validated in an in vivo model in a series of animal experiments involving left atrial injection of albumin coated air microhuhbles (Alhunex®). Videointensity curves obtained during contrast enhancement of the myocardium were analyzed to provide values of regional myocardial blood flow (in mL/min]100 g) in 45 myocardial regions of interest defined in 7 experiments performed on 4 animals. The values obtained with our approach correlated well (r = 0.77, p < 0.001) with standard reference measurements based on radiolabeled microspberes. The interteclmique variability was found to be smaller than the intersegment variability characterizing our technique. The difference between the mean flow values obtained with microspheres for segments of the entire heart and the mean flow obtained with our technique for all regions of interest ranged between 1 to 19% in the 7 experiments. In its present form, based on left atrial or left ventricnlar injection of contrast solution, this method may allow, for the first time, quantitative evaluation of myocardial regional blood supply in the cardiac catheterization laboratory or the operation theater. With further development of contrast agents suitable for transpulmonary enhancement of the myocardium by peripheral intravenous injection, this technique may provide the basis for the noninvasive quantitative measurement of myocardial tissue blood flow.

Key Words: Myocardial perfusion, Regional tissue blood flow, Echocardiographic contrast media, Indicator dilution, Mean transit time.

gions of the heart muscle (Braunwald and Sobel 1984). Reduced blood flow in the coronary capillary bed leads to insufficient blood supply and causes impairment of myocardial function. To date, there is no readily available method for the quantitative assessment of myocardial tissue blood flow, either invasive or noninvasive (Armstrong 1986; Feinstein 1986; Marcus et al. 1987; Martin 1989). Most of the current techniques allow only the assessment of relative changes in regional myocardial blood flow (Marcus et

INTRODUCTION The need for fast, reliable and noninvasive methods for the early detection of coronary heart disease may hardly be overestimated. The most important parameters in the assessment of the physiological impact of coronary disease is the blood supply to various reAddresscorrespondenceto: Prof. S. Akselrod,Schoolof Physics, Tel AvivUniversity, Tel Aviv 69978, Israel. 619

620

Ultrasound in Medicineand Biology

al. 1987), and none is sufficiently sensitive and accurate to supply a reliable absolute quantitative measurement in vivo. It is agreed that a reliable method for the quantitation of regional myocardial perfusion would provide the means for early detection of myocardial ischemia and possibly lessen the threat of coronary heart diseases (Armstrong 1986; Feinstein 1986; Marcus et al. 1987; Martin 1989). One of the developments which has the potential to satisfy the clinical need for direct assessment of regional myocardial blood flow is associated with the use of contrast materials in ultrasonic imaging of the heart (Armstrong 1986; Braunwald and Sobel 1984; Feinstein 1986; Feinstein and Shah 1986; Feinstein et al. 1986; Marcus et al. 1987; Martin 1989). Different solutions have been suggested as carriers of relatively stable reproducible microscopic air bubbles comparable in size with the red blood cells (about 5 #m in diameter), and therefore suitable for echocardiographic enhancement in humans (Berwing and Schlepper 1988; Cheirifet al. 1988; Keller et al. 1986, 1989; Reisner et al. 1989; Rovai et al. 1987; Vanderberg et al. 1989). An indirect qualitative assessment of myocardial tissue blood flow has been achieved by visual examination of the differences in time dependence of the ultrasonic intensity reflected from various regions of the myocardium, following intracoronary or aortic root injection of contrast media (Armstrong et al. 1982, 1983; Feinstein et al. 1988; Kaul et al. 1984, 1987a; Kemper et al. 1985; Lang et al. 1986; Tei et al. 1983). This allowed the differentiation between underperfused and normally perfused myocardial tissue. There have been several suggestions for providing quantitative markers related to regional blood flow, such as the peak gray level (Feinstein and Shah 1986; Vanderberg et al. 1988), the time of peak contrast (Reisner et al. 1989), the width of the contrast appearance and washout curves (Keller et al. 1988), the area under these curves (Keller et al. 1988a), the time constants of best fit gamma-variate functions (Kaul et al. 1989; Shapiro et al. 1990a), etc. However, a method for direct quantitative measurement of regional blood flow based on contrast echocardiography has yet to be presented. The aim of our study is the in vivo quantitation of regional myocardial perfusion by the analysis ofechocardiographic data acquired during contrast enhancement of the myocardium. This enhancement is achieved by injecting a contrast solution at a distant site (the left atrial cavity in the present study). Our approach was designed to be independent of the configuration of the contrast inflow reaching myocardial tissue, as well as on the dose of contrast solution.

Volume 19, Number 8, 1993 BASIC PRINCIPLES Assumptions 1. The intravascular volume fraction fwithin a specific myocardial tissue sample is assumed to be constant during the time of measurement, apart from the intracyclic variations induced by cardiac contractions. 2. Total blood flow through the tissue sample is assumed to be invariant during measurement time. This assumption may be considered fairly true as measurement time is made short enough. 3. A nearly linear relation is assumed between the echocardiographic videointensity and the concentration of the bubbles in a sample volume. This is a very rough approximation for several reasons. First, the relation between concentration and backscatter intensity may be considered nearly linear only at low concentrations (Powsner et al. 1986). Also, nonlinear gray maps and dynamic range compression are widely used in the ultrasound equipment. Besides, as long as the time-gain compensation (TGC) function of the echo equipment is insensitive to variations in the acoustic properties of the structures more proximal to the transducer, it is not even evident that local variations in videointensity reflect local variations in the acoustic properties of the corresponding tissues and organs. 4. Tracer inflow reaching the sample is not assumed to be instantaneous. This assumption gives our approach its greatest advantage, since the ultimate goal is to allow the measurement of regional myocardial perfusion using peripheral intravenous injection of contrast material. A more careful and detailed discussion of the above assumptions is provided in the Discussion section below. However, already at this point, we would like to stress that the severity of the difficulties posed by these assumptions should not be underestimated. Indicator dilution The approach is based on the application of the basic principles of indicator dilution theory (Gonzalez-Fernandez 1962; Meier and Zierler 1954; Trautman and Newbower 1984; Zeirler 1958, 1962) to flow dynamics in the microvasculature of a selected area of myocardial tissue. According to this theory, the amount of fluid flowing through a hydrodynamic system of a known volume in unit time (i.e., the flow) may be evaluated by introducing into the system

Myocardial perfusion measurement by contrast echocardiography • V. MOR-AVIel al.

small inert particles that can be traced by appropriate means. The dynamics of the ideal tracer must be identical to those of the fluid in the system, which was shown to be a good approximation in the case of sonicated Albumin microbubbles (Keller et al. 1989). Thus, the mean transit time of the tracer particles may represent the time for the entire volume to be replaced by the inflow of fresh fluid. Therefore, provided the volume of fluid in the system and the mean time of transit of the tracer particles through the system are known, the flow equals the ratio between these two parameters (Gonzalez-Fernandez 1962; Meier and Zieder 1954; Trautman and Newbower 1984; Zeirler 1958, 1962). Thus, in a tissue sample of volume V, containing microvasculature of volume iV, the knowledge of the mean transit time of the tracer, (t), allows to obtain the regional flow in the sample, J:

J - fV (t)

"

(1)

Evaluation of the mean transit time The evaluation of the mean transit time, (t), under conditions of noninstantaneous hand-injection of contrast solution, has been achieved by frequency domain deconvolution of the characteristic response function. We have described the procedure in detail elsewhere (Mor-Avi et al. 1993). The computational aspects of this technique were validated by an extensive computer simulation study, and the optimization of the related parameters was achieved. The procedure was shown to yield reliable values of the mean transit time, (t), in a wide range of contrast inflow configurations and noise levels.

Evaluation of the blood volume Blood volume in the microscopic blood vessels supplying the sample tissue may be evaluated by the analysis of the time dependence of echo intensity curves obtained from the different regions of the heart. The amount of tracer, Mp(t), in volume V of blood in a homogeneous blood pool upstream of the myocardial region of interest (such as the left ventricular cavity or the aortic root), may be expressed as:

M~(t) = Vep(t).

(2)

Similarly, the amount of tracer in the tissue sample, Ms(t), may be expressed as the product of blood volume of the capillary bed of the sample tissue,f V, and the concentration of the tracer in the capillaries, cs(t):

621

Ms(t) = fVG(t).

(3)

Dividing eqn (3) by eqn (2) and integrating with respect to time we obtain:

f~+~ooMs(t)dt _ f I

+~ G(t)dt-

f_~ Mp(t)dt

(4)

cp(t)dt

It should be noted that the tracer concentration, cs(t), reaches its maximum value with some delay relative to the concentration in the pool, cp(t). Also, due to the different pathways used by the tracer particles in the sample tissue, a certain mixing takes place. Thus, the maximal tracer concentration in the blood entering the sample tissue, Cs(t), does not reach the same level as that in the reference region, cp(t). However, since bubbles flow with blood from the left ventricular cavity to the myocardium, the concentration of bubbles per unit volume of blood reaching myocardial tissue is similar to that in blood leaving the left ventricular cavity. Therefore, the conservation of mass requires that:

G(t)dt = oo

cp(t)dt;

(5)

oO

and therefore eqn (4) is immediately reduced to:

f =

f•o•Ms(t)dt ~o

(6)

f f ooMp(t)dt oo

The result of eqn (6) promises to provide an estimate off(Fig. 1). However, Ms(t) and Mp(t) are not directly measurable, and they must be replaced by the ultrasonic reflection intensities, Is(t) and Ip(t) as explained below. As already mentioned above, we assume a linear relationship between tracer-produced intensity of the ultrasonic reflection from a certain region of interest and the amount oftracer per unit volume in that specific region. The validity of this approximation is discussed later on. This assumption allows us, however, to express the intensity of the reflection from the blood pool, Ip(t), as:

Mp(t) . Ip(t)= ~, V '

622

Ultrasound in Medicine and Biology

Volume 19, Number 8, 1993

cal: in

bubbleS in blood

umu

$

S

= i n t r a v a s c u l a r v o l u m e fraction

Fig. 1. The media in the two regions of interest shown in this figure are different (see magnifications on the left), and hence the difference between the intensity curves (on the right).

or

VI.(t) Mp(t)=

~1

(7)

Similarly, the reflection intensity from the myocardial tissue sample, Is(t), may be expressed as:

Therefore, measuring tracer-produced echo intensities in the myocardial tissue sample and the left ventricle or the aortic root may provide us with an estimate of the fraction of volume of the sample occupied by capillary bed.

Evaluation of regional perfusion

M,(t) . 6 ( 0 = ~2

V

According to eqn (1), knowing the blood volume,

'

or

f V , contained in the selected region of the myocardial (8)

M~(t) =

VI~(t)

Though the constants, ~'~and g'2,may not be equal for two different regions of interest located at different depths, we may assume that their values are similar (fl -~ ~'2),to the extent that videointensity reflects regional acoustic properties. Dividing eqn (8) by eqn (7) and integrating with respect to time, we obtain:

+= M~(t)dt

I~(t)dt (9)

oo

tissue (in mL) and the mean transit time, (t), of the tracer (in min), we obtain the regional blood flow in mL/min. In order to obtain regional perfusion in mL/ min/100 g, the result of eqn (1) must be divided by tissue mass (in g) and multiplied by 100. Assuming that characteristic density of mass of soft living tissues may be approximated to 1 g/cm 3, this yields the following:

f V 100

J= }

lOOf

v-t = %5-

(ll)

Substituting eqn (10) for f, we may obtain an estimate for myocardial regional perfusion in units of mL/ min/1 O0 g:

oo

Equating (9) with (6) we obtain:

f

j =

_ ~ I~(Odt

f~_

=

f ~o~ /p(t)dt oo

(lO)

~

oo oo

I~(t)dt lO0

f••

I,,(t)dt ( t )

oo

(12) "

Myocardial perfusion measurement by contrast echocardiography • V. MOR-AWet al.

METHODS The experimentation was carried out in two phases. Phase I experiments were conducted on 14 mongrel dogs (weight, 17-23 kg). These experiments were aimed at the optimization of the measurement conditions, including the handling of the contrast agent, the dosage of contrast solution and its composition, the preferred site and rate of injection, the ultrasonic imaging parameters (such as acoustic power, gain, TGC, transducer location and orientation), breathing conditions, definition of regions of interest, etc. Phase II experiments were designed for the validation of our echo contrast technique with the standard reference technique based on radiolabeled microspheres. Seven phase II experiments were performed in 4 animals (weight, 19-22 kg) according to the conditions defined by the observations and conclusions in phase I experiments.

Experimental techniques A diagrammatic depiction of the experimental setup is presented in Fig. 2. The animal was anesthetized with alpha chloralose (0.5 mg/kg body weight,

623

intravenously), which was added as required throughout the experiment. Femoral arterial blood pressure and electrocardiogram were continuously monitored. The animal was intubated and ventilated with a Bird Mark-8 respirator (Palm Springs, CA). Body temperature was kept constant at 37°C +_ 2°C by a heating lamp. A polyethylene catheter was inserted into a femoral vein to sustain an open intravenous line. Sternotomy and pericardiotomy were carried out, and a pericardial cradle was established for suspension of the heart. An additional 5 F catheter was introduced into the left atrial cavity through the left atrial appendage.

Contrast injection Albunex ® microbubbles (Molecular Biosystems Inc., San Diego, CA) were used for echocardiographic enhancement. Albunex ® vials were slightly hand-agitated prior to the injection and vented with an 18gauge needle, with its tip positioned in the air space in the vial. Another 18-gauge needle was used to transfer carefully and slowly 0.2 to 0.3 mL ofAlbunex ®into 1 mL syringe. Albunex ® was subsequently diluted in a 5% human albumin saline solution to produce 3 mL of the contrast solution used for each contrast injection. Manual bolus injections (over 10 cardiac cycles) were performed via the left atrial catheter.

Ultrasonic imaging

¥ I~'~o s~ng unit ¥

I

ideo stgnld: 2 ~lim c r o s t - s e c t k ~ a l i m . g e s ~ ' ] e! intenshy proportional to echo intens~ty)J

~.nd_

SC~r'O*ted ~1 dia Iltolie f r m s . . J - -

¥

V . . . . . ---I *"~y ~ t ~ Image monitor

Ik f r m

__---7--]

v~,o-~.,. video-prlnter

I

]

grabber b o a r d

T

ff oo.rd,-, "o a' 7 Fig. 2. Schematic representation of the experimental setup.

Echocardiographic two-dimensional short axis images of the left ventricle were obtained during contrast enhancement, using the Aloka 870-SSD ultrasound apparatus with a 5 MHz phased-array transducer. The transducer was fixed on the right ventricular epicardial surface avoiding undue pressure on the right ventricular free wall. A small amount of aqueous gel was introduced into a surgical glove covering the transducer, in order to increase the distance between the transducer and the heart, and thus place the entire left ventricle within the view angle of the transducer. Acoustic power was set as low as possible, while retaining the good quality of the images. All manual TGC controls were set off, and TGC was performed using the default setting of the TGC control of the machine. The overall gain was adjusted to the dynamic range of the frame grabber, so as to produce a gray-scale value close to the maximum of the scale (255) at peak enhancement of the left ventricular cavity. This is essential in order to achieve maximal sensitivity of the image processing system to minor changes produced by contrast injection in the myocardium. All these settings were kept constant throughout the entire experiment.

Ultrasound in Medicineand Biology

624

Volume 19, Number8, 1993

Data acquisition

t2

The respiration was stopped for the short period of image acquisition (64 heartbeats, less than l min), in order to minimize cardiac translation due to respiratory chest movements. A series of end-diastolic images of 64 successive heart beats was selectively acquired for each contrast injection by a cine-memory of the ultrasonic scanner using ECG gating. These included l 0 preinjection frames, followed by additional 54 postinjection frames. All images were transferred to an AT compatible personal computer equipped with a frame grabber (DT-2851 model, Data Translation Co., Marlboro, MA) for further analysis.

Experimental protocols Phase I experiments. The aim of these experiments was to acquire the skills necessary for performing the validation of the measurement of regional flow in phase II. Each of these experiments was designed for a specific purpose, and each was conducted according to a predetermined protocol, different from one experiment to another. This extensive learning period determined the conditions for performing phase II. Phase H experiments. Each experiment included three injections of contrast solution. During each injection, a series of short-axis view images of a single left ventricular slice was obtained at one of the three levels 2, 3 or 4 (see Fig. 3). Subsequently, radiolabeled microspheres were injected to complete the experiment and allow validation of the measurement (see "'Validation by microspheres" below).

64 ,/'

2

2 3 4 5

3

4 r

/

5 i

Lv slice 3

11II

Fig. 3. For the echocardiographic measurements of regional perfusion, the left ventricular mass was virtually divided into 5 slicescorresponding to a short axis echocardiographic view at different levels. A series of 64 images was obtained for each of the slices2, 3 and 4 during contrast enhancement of the myocardium. The figure shows a schematic representation of data acquisition from slice 3.

d --O 2

Fig. 4. For the validation of regional perfusion measurements by radiolabeled microspheres, the left ventricular mass was cut into 5 slices, roughly corresponding to the echocardiographic slices in Fig. 3. Each slice was further divided into wedge-shapedpieces. Each piece was subjected to gamma spectroscopy, necessary for the evaluation of blood supply to this specific piece.

Postprocessing The echocardiographic images were processed as follows, using a specially designed computer software: 1. For each injection, a reference region within the left ventricular cavity, and several myocardial regions of interest were selected using a mouse-driven cursor. These myocardial regions corresponded roughly to the wedge-shaped pieces described in Fig. 4 (endocardium and epicardium were, however, excluded). 2. The mean videointensity within each region of interest and the reference region was evaluated in each of the 64 consecutive images. Thus, for each region, a curve was obtained representing regional intensity as a function of time. 3. The intensities obtained for the heartbeats prior to contrast injection were averaged over all preinjection frames, so as to obtain the background intensity. Each background value was then subtracted from the corresponding curve, resulting in a curve representing exclusively tracer-produced regional intensity. 4. The mean transit time of the contrast through each region of interest was evaluated by frequency domain deconvolution analysis of the impulse response function corresponding to that specific region, according to an algorithm described elsewhere (Mor-Avi et al. 1993) in detail. 5. The integrals in eqn (10) were obtained by the summation of the values of videointensity in the region of interest and the reference region.

Myocardialperfusion measurement by contrast echocardiography• V. MOR-AVlet al. 6. Regional perfusion was finally computed from the results of the latter two phases according to eqn (12).

Validation by radiolabeled microspheres Seven experiments, including measurements of regional myocardial blood flow by the standard reference technique (Heymann et al. 1977; Kaul et al. 1989; Kerber et al. 1975; Marcus et al. 1975), based on radiolabeled microspheres were performed. In each experiment, microspheres (10-12 #m diameter) labeled with 46Sc (889.2 keV and 1120.5 keV), or SSSr (514.0 keV) isotopes (Dupont, New England Nuclear Co.) were injected following the acquisition of echocardiographic data. During each injection of microspheres, a reference arterial blood sample was obtained for 60 s, using a withdrawal pump (Harvard, Model 644) with a 8 F catheter placed in the right femoral artery. Withdrawal of reference blood was initiated just prior to injection of the microspheres, and the withdrawal rate was determined by weighing the collected reference blood. At the end of each experiment, the animal was sacrificed. The heart was excised and the left ventricular myocardium was separated from other tissues. It was subsequently cut into 5 horizontal slices about 1 cm thick. Each slice was divided into 5 to 6 wedgeshaped segments. Each segment was sealed in a 20 mL plastic vial for activity counting (Fig. 4). All tissue samples were weighed immediately upon sealing. Reference blood was transferred into separate vials, including the saline used for the thorough flushing of the syringe, so that each vial contained up to 10 mL of fluid. The activity of each wedge-shaped myocardial tissue sample and reference blood sample was measured using a high resolution Ge detector and a multichannel analyzer PC board, operated by a commercial software package (System-100, Canberra Industries Inc., Meriden, CT). Spectral resolution was set so as to produce complete separation between the above three spectral peaks characteristic to the gamma emission spectra of the above isotopes. Each sample was counted, so as to produce statistical error of not more than 15%. Count rates were obtained for each sample by dividing total counts by counting time. For each isotope, count rate of the reference blood was evaluated by the summation of the count rates of all blood vials. Blood flow to each tissue sample (in mL/min) was evaluated as follows: tissue CR J - reference CR WR,

(13)

625

where CR denotes count rate, and WR stands for the withdrawal rate (in mL/min). Regional perfusion, j, in each tissue sample (in mL/min/100 g) was evaluated by dividing the flow, J, by the sample mass (in g) and multiplying by 100.

Statistical analysis The correlation between the values of regional myocardial blood flow obtained by our method for the different regions of interest and with microspheres for the different myocardial segments was evaluated by computing the Pearson product moment correlation coefficient (Glantz 1981). The hypothesis that there is no trend relating these two variables was tested using the t test (Glantz 1981). RESULTS

Phase I experiments: Evaluation of conditions Phase I experiments resulted in the determination of the conditions (handling of contrast agent and its dose, respiratory movements, selection of regions of interest, etc.) for the optimization of the assessment of regional myocardial perfusion. Some of the observations made during these experiments were previously reported by other investigators. Therefore, we briefly present below only those results necessary to justify the experimental techniques used in phase II experiments. Contrast enhancement. Figures 5(A-D) illustrate four examples of a short axis image obtained under control conditions: (A) prior to contrast injection; (B) immediately after contrast injection (notice the enhanced contrast of the left ventricular cavity); (C) a few heart beats later [notice that echo intensity of the myocardial tissue is slightly enhanced as compared with image (A)]; (D) following contrast washout. Intensity curves. A typical example of contrast enhancement in a myocardial region of interest and the left ventricular cavity is shown in Fig. 6. It should be noted that myocardial enhancement is considerably lower as compared with that in the left ventrieular cavity (see scaling of the myocardial curve in Fig. 6). The left ventricular enhancement starts with the initiation of contrast injection. The intensity in the myocardium also increases; however, this increase occurs one to two heart beats later. Following washout of contrast from the left ventricle, myocardial enhancement is barely distinguishable above the noise level, and the entire myocardial curve seems in this case narrower than the left ventricular curve.

626

Ultrasound in Medicineand Biology

Volume19, Number 8, 1993

Fig. 5. Four examples of the images obtained as in Fig. 3 (for detailed description, see text).

found that venting a vial with an additional needle, according to the manufacturer's instructions, eliminates this degradation.

Contrast agent. Figure 7 shows the effect of bubble disintegration caused by negative pressures, induced in the Albunex ® vial during solution withdrawal by syringe without venting the vial. The two curves show videointensity, averaged over the left ventricular cavity and integrated with respect to time, following injections of increasing concentration of the contrast agent. The upper curve represents the integrated intensity obtained using for each concentration a new Albunex ® vial. The lower curve represents similar data obtained using contrast agent from the same vial 5 min later. A significant decrease in the left ventricular enhancement is observed following 5 min of negative pressure exerted on the bubbles. It was

Albunex* dose. Increasing amounts ofAlbunex ® above 30% in a 3 m L albumin solution bolus did not further enhance the integrated intensity (Fig. 7). It was thus understood that doses below 1 m L should be used to permit the approximation of the linear relation between the concentration of bubbles and the excess videointensity. Fortunately, amounts of AIbunex as low as 0.2 m L were found to produce sufficient enhancement of the myocardial tissue, above the noise level.

200 LV cavity

>.,

, (xl O) tissue region

150

¢¢.. m ~

100

0 "0 , m

>

50

i

0

i

32 '

16 heart

'

48

'

64

beat

Fig. 6. An example of contrast enhancement of a myocardial re#on of interest and a reference re#on defined in the left ventricular cavity, following left atrial injection of contrast solution. Both curves were brought to the same scale by multiplying the myocardial videointensities by 10.

Myocardial perfusion measurement by contrast echocardiography • V. MOR-AVIet al.

627

3000' ../.-'°"

t¢-

2000

10 /t.o o''

°~,e..~"~°"

looo

n e w vial

r-

5 rain

m

0

o

o15 i 1.'5 Albunex close (ml)

I

Fig. 7. The effect of bubble degradation induced by negative pressure created in the Albunex vial by solution withdrawal by syringe without venting. The y-axis presents videointensity, measured in the left ventricular cavity following injections of increasing concentration of Albunex, and integrated with respect to time, so as to represent the total amount of contrast traversing the left ventricle. The upper trace was obtained using a new vial for each injection. The lower trace was obtained by injecting each concentration from the vial used for the previous injection 5 min earlier.

Distal acoustic shading. Higher doses of A1bunex ®caused significant shading of the distal area of the images during the contrast enhancement of the left ventricular cavity (Fig. 5B). Therefore, it is crucial to minimize the shading by choosing adequate doses of contrast material. Moreover, the shading is nonlinear, since each pixel is assigned a gray level according to the reflections along the path of the acoustic beam propagating between the transducer and the corresponding point. The reflections depend on the acoustic properties of tissues and structures that the beam traverses, which are different for each path. Also, the reflections, and, subsequently, the shading, vary with time, while the bolus of contrast material passes through the cardiac chambers and the myocardium. Phase H experiments: Measurements of regional perfusion The results of the measurements of regional myocardial blood flow obtained in the 7 experiments involving radiolabeled microspheres are presented in Table 1. Experiments 1 and 6 included a single echo contrast injection and single ultrasonic slice imaging. In each of the other five experiments, up to three slices were subjected to perfusion measurement based on echo contrast enhancement (Fig. 3). For each slice, three regions of interest were chosen, where no sig-

nificant acoustic shading occurred during contrast enhancement of the left ventricle. These regions of interest were subjected to the above detailed videodensitometric analysis of the transitory contrast enhancement. The results of the regional blood flow measurements obtained from echo contrast data for each region of interest were compared with those obtained with the radiolabeled microspheres for the corresponding myocardial segment (Table 1). In each of the 7 experiments, flow measurements by the two techniques showed highly comparable results. The exception to this were two segments in experiment 2: segment 3--slice 2, and segment 3--slice 3 (Table 1), both deviating significantly from the reference values. It should also be noted that flow values obtained by both contrast echo data and microspheres varied considerably between the different experiments, including those conducted in the same animal (see experiments 2 and 3). The mean flow values obtained for all segments in each experiment (Fig. 8) allow the comparison between the average myocardial flow measured by our method versus the microspheres. The intertechnique variability in all experiments (columns 4 and 6) is smaller than the corresponding standard deviation (SD) characterizing the intersegment variability in

Ultrasound in Medicine and Biology

628

Table 1. The results of regional blood flow measurements (in m L / m i n / l O 0 g) obtained from echo contrast data for each region of interest and those obtained with the radio|abeled microspheres for the corresponding myocardial segments. Exp.

Dog

Label

Slice

1

15

Sc

4

2

16

Sc

2 3

4

3

4

16

17

Sr

Sc

3

2

3 4

5

17

Sr

2

Seg.

Flow

Err.

CE flow

1 2 3 4 5

73 58 59 63 77

12 9 9 I0

84 53 78 68

11

111

1 2 3 1 2 3 1 2 3

169 186 155 177 165 152 185 146 168

25 29 23 26 25 24 28 25 26

183 164 81 145 144 216 163 142 180 119 77 89

1

112

2 3

112 114

35 34 34

1 2 3 1 2 3 1 2 3

110 102 79 100 88 95 107 104 115

13 13 10 12 12 12 13 13 14

130 87 51 117 92 88 151 84 138

1

18 19 22 19 19 18

109 129 144 126

2 3

112 119 144 128 126 115

112 103 121 140

2 3

1

6

7

DI

DI

Sr

Sc

2

2

3 4

1

105

7

2 3 4

103 128 143

8 10

1 2 3 1 2 3 1 2 3

85 91 97 69 94 104 92 113 98

5 5 6 5 7 7 6 8 7

11

111 100

93 102 139 55

111

Volume 19, Number 8, 1993

Figure 9 presents the correlation between the regional tissue blood flow (in mL/min/100 g) obtained using contrast echo for all 45 segments in Table 1 according to our algorithm, against those measured with the microspheres. The correlation coefficient for the two sets of data was found to b e r = 0.77, corresponding to p < 0.001. DISCUSSION To date, the existing methods for the assessment of impaired tissue blood supply due to coronary heart disease are focused on either the pathological changes in the morphology of the large coronary arteries or on a qualitative evaluation of blood flow by radionuclide imaging or other techniques. It is agreed today that the severity of coronary artery occlusion is an insufficient marker of the extent of reduction in tissue blood supply. It is well known that collateral circulation may in many cases supply the vital needs of the tissue via the alternative network of blood vessels (Kaul et al. 1987b, 1991; Widimsky et al. 1988). A measurement of regional tissue blood flow would provide the most direct quantitative information about the blood supply to any myocardial segment, and thereby the extent and the physiological importance of the ischemic process. Contrast echocardiography is increasingly referred to as having the potential ability to address the clinical need for the quantitative evaluation of regional myocardial tissue blood flow (Armstrong et al. 1982, 1983; Armstrong 1986; Berwing and Schlepper 1988; Cheirif ct al. 1988; Feinstein 1986; Feinstein

250 o

"2

~150 Eo~lO0

121 96 119

I.I. "1=1

100

m

Abbreviations: Exp. = experiment; Seg. = segment; Err. = error; CE = contrast echocardiographic.

our method (column 4). Variations in mean heart rate and mean blood pressure should be noted between experiments 2 and 3, experiments 4 and 5, and e x p e r i m e n t s 6 and 7, performed on the same three animals correspondingly.

200

_~

so

1

2

3

4 5 Experiment

6

7

[ ] Microspheres [ ] Contrast echo Fig. 8. The mean flow values (in mL/min/100 g) obtained for all regions of interest in each experiment by our method, and the average myocardial flow measured in all corresponding segments by the microspheres technique. Standard deviations are also shown.

Myocardial perfusion measurement by contrast echocardiography • V. MOR-Avl et al.

629

250 0

200 ....//~

"~150

..,.:"

m

../~"

50

~

Q

......~ _

B

8

..I~"

D

r=0.77 p
la

....'"

0

I

0

25

I

50

1

I

75

I

I

I

I

I

11~ 125 150 175 2110 225 250

mierospheres Fig. 9. Values of regional tissue blood flow (in mL/min/100 g) obtained by the analysis of echo contrast appearance and washout curves in 45 different regions of interest in 7 experiments, against those measured with radiolabeled microspheres in the corresponding myocardial segments (r = 0.77, p < 0.001). The dashed diagonal line represents the hypothetical equality between the two measurements.

and Shah 1986; Feinstein et al. 1986, 1988; Kaul et al. 1984, 1987a, 1989, Keller et al. 1986, 1988a, 1988b, 1989; Kemper et al. 1985; Lang et al. 1986; Marcus et al. 1987; Martin 1989; Meerbaum 1991; Reisner et al. 1989; Rovai et al. 1987; Tei et al. 1983; Vanderberg et al. 1988, 1989). However, an absolute quantitative measurement of regional perfusion based on contrast echocardiography has yet to be presented (Kaul 1991; Meerbaum 1991). The aim of this study was to devise an algorithm for the quantitative evaluation of regional myocardial blood flow by analyzing echocardiographic contrast appearance and washout curves. Indicator dilution techniques generally require the knowledge of the injected amount of contrast material. Other investigators proposed standardization of contrast injection by using a power injector (Vanderberg et al. 1988). However, unless intracoronary injection is used, a specific standard form of contrast inflow to the myocardium cannot be warranted, since volume of dispersion largely depends upon physiological factors, such as cardiac output. Our aim was to develop an algorithm, which would be primarily independent of the dose of contrast material and the rate ofits injection. Indeed, one of the main advantages of our approach is that the amount of contrast material and injection rate are important only to assure adequate contrast enhancement of the myocardium. This simplifies the measurement procedure and lessens its chances to be jeopardized by interindividual variability. In addition, it is essential for the assessment of regional myocardial blood flow using periph-

eral intravenous injection, involving transpulmonary passage of contrast agent. Recent publications (Cheirif et al. 1988; Feinstein et al. 1990; Keller et al. 1987; Rayleigh 1945; Shapiro et al. 1989; Smith et al. 1989; Villanueva et al. 1992) confirm the feasibility of such noninvasive contrast enhancement of the myocardium.

Validity of the theoretical approach The theoretical model is based on several assumptions associated with (a) stability of the intravascular volume during measurement, (b) size uniformity and stability of the microbubbles, and (c) the relation between bubble concentration and videointensity. The validity of these assumptions within the framework of this study should be carefully examined.

Stability ofintravascular volume. It may be questioned whether intracyclic variations in the intravascular volume comply with the requirement that this volume must not change, so as to allow its evaluation, necessary for the measurement of regional perfusion. Our approach estimates the intravascular volume as a mean value for all phases of the cardiac cycle. It is indeed essential that this mean value be constant during measurement time. However, measurement time is rather short (less than 1 min), which assures that no significant change in regional perfusion occurs during measurement.

630

Ultrasound in Medicine and Biology

Pressure variations and videointensity. The diameter of the bubbles is strongly affected by pressure variations (Fanelli et al. 1984; Gazanhes et al. 1984; Morse and Ingard 1968; Nishi 1975; Powsner et al. 1986). The intensity of the ultrasonic reflection is in turn strongly dependent on the bubble size (Powsner et al. 1986). In our method, the comparisons are made during the end-diastole. At this phase of the cardiac cycle, the ambient pressure exerted upon the bubbles in the tissue and in the left ventricular cavity is approximately equal and relatively low. Thus, this comparison may be assumed to be reliable, assuming that the fraction of bubbles that collapse under left ventricular pressures is small. Bubble stability. The reproducibility of the bubble size under pressure variations should still be considered. The physics of a bubble in liquid environment and its interaction with ultrasonic waves are very complex (Morse and Ingard 1968; Rayleigh 1945). To date, there is no theory available, which would facilitate the modeling of the effects of systemic arterial pressure on the bubble stability. There is some evidence that pressure may cause sonicated Albumin bubbles to collapse during their transit through the cardiac chambers (Shapiro et al. 1990b), and it has been suggested that bubbles may collapse under high pressure by gas diffusion through the albumin coating (Powsner et al. 1986). This issue has yet to be further investigated. Our preliminary results show that A1bunex bubbles are more stable and can better tolerate typical left ventricular pressures than those described by Shapiro et al. (1990b). Bubble concentration and videointensity. Our method assumes a linear relation between the concentration of the contrast agent and regional videointensity. This assumption implies that regional videointensity be entirely dependent on the local acoustic properties. In the vast majority of commercial ultrasonic imaging systems, this assumption is not justified. Propagation effects, such as energy losses from the beam due to scattering and different absorption mechanisms (Jaffe and Harris 1980; LaFollette and Ziskin 1986; O'Donnell 1983), are compensated for, based on a presumably uniform beam attenuation throughout the investigated tissues and organs. This TGC does not impede the diagnostic distinctions made by trained personnel. Additional manual TGC may, in some cases, even facilitate visual examination of single echocardiographic images. However, reliable quantitation of myocardial contrast enhancement requires more accurate path-dependent compensation (Mor-Avi et al. 1992). The uniformity assumption is in particular inade-

Volume 19, Number 8, 1993

quate when contrast media are involved. Regional videointensity may be strongly affected by the acoustic properties of the adjacent regions, especially in the distal areas prone to massive acoustic shading during contrast transit through the left ventricular cavity. It can, thus, be understood that without proper compensation for beam attenuation, quantitative echocardiographic measurements may be misleading (Meltzer et al. 1980). This problem has yet to be solved. In this study, distal regions of interest were excluded. We believe that repeated imaging from a different view angle or the creation of compound images may provide the solution to this problem. Alternatively, a path-dependent compensation for beam attenuation (Melton and Skorton 1983; Pincu et al. 1986; Pye et al. 1992) might be efficient. Furthermore, even if perfect compensation is possible, videointensity is usually proportional to the logarithm of the ultrasonic echo intensity, depending on the specific compression characteristic used. This issue is fundamental to the application of linear system theory, which is at the basis of our approach to the evaluation of the mean transit time. However, this problem may be easily solved by replacing the linearity assumption by the specific logarithmic dependence used in a particular machine. It is reasonable to expect that such an improvement may further enhance the reliability of our method. Moreover, it is one of the basic principles of indicator dilution that in a multiple branching hydrodynamic system, the curve reflecting output concentration of tracer as a function of time is wider than that corresponding to the input concentration. In our case, it means that contrast particles spend more time in the capillary network in myocardial tissue sample than in the left ventricular cavity. Therefore, it might be expected that myocardial enhancement last longer, or at least the same time, as that in the left ventricular cavity. However, our experimental data, as well as the observations made by other investigators (Villanueva et al. 1992), indicate that it is very difficult to detect contrast enhancement of the myocardium following contrast washout from the left ventricular cavity (Fig. 6). This apparent shorter output function is probably related to the fact that the number of bubbles in the myocardium falls below the sensing threshold of the transducer at all times other than when the left ventricular cavity still contains considerable amount of contrast. In this sense, echocardiographic contrast cannot be regarded as an ideal indicator, and the effects of this deviation from the basic requirements of indicator dilution are very difficult to appreciate. Therefore, experimental validation of the above theoretical approach is of utmost importance, and the entire ap-

Myocardial perfusion measurement by contrast echocardiography • V. MoR-Avl et al.

proach should be judged based on the experimental results.

Experimental validation Radiolabeled microspheres provide a "gold standard" method for experimental validation of measurements of regional tissue blood flow (Heymann et al. 1977; Kerber et al. 1975; Marcus et al. 1975). Regional perfusion can be determined with very high precision in a series of consecutive states, using different radioisotopes. However, this technique requires the experimental animal to be sacrificed. The basis of our approach was to use a homogeneous blood pool upstream of the myocardial tissue sample, such as the left ventricular cavity or the aortic root, as the reference for the evaluation of the intravascular volume fraction of the tissue sample. The contrast transit curve obtained from the above pool served also as a first approximation of the contrast inflow function reaching the sample tissue. The resuits obtained using this approximation were shown to correlate well with the microspheres measurements under conditions of this study, in spite of the limitations detailed below. The credibility of this correlation is strengthened by the low intersegment variability of the results in each experiment, both in microspheres and contrast echo methods (see SDs in Fig. 8). Even more convincing is the fact that the intertechnique variability in all experiments is smaller than the intersegment variability (Fig. 8). The variations in flow measured in two subsequent experiments performed on the same animal reflect the hemodynamic changes between experiments, as evidenced by changes in blood pressure and heart rate (Fig. 8).

Limitations of the experimental study Transducerfixation. For the adequate evaluation of transitory contrast enhancement in a myocardial region of interest, the transducer has to be as stable as possible to produce images of the same tissue segment in every consecutive heart beat. Such a stability is difficult to obtain from the epicardial surface of a beating heart. Though the variations in transducer location and orientation relative to the heart may be effectively reduced by firmly attaching the transducer to the epicardial surface, occasional translocation cannot be excluded. Breathing movements. Chest movements present a major interference, as the heart and the transducer are translocated through the different phases of the breathing movements. Thus, different tissue segments may be scanned during the different phases of

631

the breathing cycle. For this reason, in this study the respirator was stopped during the image acquisition. In the future use in the clinical setup, the number of sampled heart beats will have to be reduced from 64 to 32 to minimize breathing effects.

Regions of interest. One of the major difficulties was the choice of fixed regions of interest. Sometimes, even a minor displacement of the heart during image acquisition may introduce very high amplitude noise, due to occasional overlapping between the region of interest and the adjacent highly echogenic structures, such as the endo- and epicardium. In this study, shifting the regions of interest according to the movement of a specific, interactively marked, distinct echocardiographic detail, was performed in each of the consecutive images. An automatic tracking algorithm would probably efficiently resolve this difficulty. Anatomy and echocardiography. It is important to understand that a full alignment between the echocardiographic regions of interest superimposed on the images of a beating heart and the myocardial segments used for gamma spectroscopy is extremely difficult to achieve. T.he alignment was based on the myocardial anatomy, which permits only a rather rough overlap. The regions of interest used in our study could be described as subsamples of the physical wedge-shaped segments. However, homogeneous flow was assumed during each experiment, and therefore accurate alignment was not crucial. Nevertheless, the cross validation of our measurements by radiolabeled microspheres resulted in very good correlation. Summary To summarize, the positive results of the present study provide strong evidence that regional myocardial tissue blood flow can be measured quantitatively based on the analysis of myocardial echo contrast enhancement using our algorithm. The results of this study are even more impressive if one considers the inherent difficulties posed by the physics of the contrast agents and technical limitations of the echocardiographic imaging techniques. Since the measurement of myocardial perfusion is undoubtedly one of the most sought after in medicine, it is of great importance for the manufacturers of both the ultrasound equipment and the contrast agents to direct the necessary efforts towards the resolution of these difficulties, so that these measurements could actually be made in the clinical setting. The need for further validation of our method under various hemodynamic conditions is obvious. Following further experimental validation, we anticipate that clinical estimation of regional myocardial blood flow using direct left ventricular in-

632

Ultrasound in Medicine and Biology

jection of contrast media will be practical. In this invasive form, the method may be a useful tool for the intraoperative evaluation of blood supply during bypass surgery (Goldman and Mindich 1987; Keller et al. 1990) or for the evaluation of the effects of coronary angioplasty (Grill et al. 1990). However, the measurement in its final noninvasive form, based on a peripheral intravenous injection of contrast material, promises to be suitable for mass screening, aimed at detecting coronary artery disease during its early asymptomatic stages. Acknowledgements--V. M. is a recipient of the Levi-Eshkol Fel-

lowship of the Israel Ministry of Science and Technology. The project was supported by the Elizabeth and Nikolas Slezak Foundation, The Abramson Center of Medical Physics, and the Ministry of Health.

REFERENCES Armstrong, W. F.; Mueler, T. M.; Kinney, E. L.; Tickner, E. G.; Dillion, J. C.; Feigenbaum, H. Assessment ofmyoeardial perfusion abnormalities with contrast enhanced two dimensional eehoeardiography. Circulation 66:166-179; 1982. Armstrong, W. F.; West, S. R.; Mueller, T. M.; Dillion, J. C.; Feigenbaum, H. Assessment of location and size of myocardial infarction with contrast enhanced echocardiography. J. Am. Coil. Cardiol. 2:63-72; 1983. Armstrong, W. F. Assessment of myocardial perfusion with Contrast enhanced echoeardiography. Echocardiography 3:355370; 1986. Berwing, K.; Schlepper, M. Echoeardiographic imaging of the left ventricle by the peripheral intravenous injection of echo contrast agent. Am. Heart J. 115:399-408; 1988. Braunwald, E.; Sobel, B. E. Coronary blood flow and myocardial ischemia. In: Braunwald, E., ed. Heart disease--A textbook of cardiovascular medicine, 2nd ed. Philadelphia: W. B. Saunders Co.; 1984:1235-1261. Cheirif, J.; Zoghbi, W. A.; Raizner, A. E.; Minor, S. T.; Winters, W. L.; Klein, M. S.; De Bauche, T. L.; Lewis, J. M.; Roberts, R.; Quinones, M. A. Assessment of myocardial perfusion in humans by contrast echocardiography. I. Evaluation of regional coronary reserve by peak contrast intensity. J. Am. Coil. Cardiol. 11:735-743; 1988. Fanelli, M.; Prosperetti, A.; Reali, M. Shape oscillations ofgas-vapour bubbles in liquids. Part I: Mathematical formulation. Acustiea 55:213-223; 1984. Feinstein, S. B. Myocardial perfusion imaging: Contrast echocardiography today and tomorrow. J. Am. Coll. Cardiol. 8:251-253; 1986. Feinstein, S. B.; Shah, P. M. Advances in contrast two-dimensional echoeardiography. Cardiovasc. Clin. 17:95-102; 1986. Feinstein, S. B.; Lang, R. M.; Dick, C.; Neuman, A.; AI-Sadir, J.; Chua, K. G.; Carrol, J.; Feldman, T.; Borow, K. M. Contrast eehocardiographic perfusion studies in humans. Am. J. Cardiac Imaging 1:29-42; 1986. Feinstein, S. B.; Lang, R. M.; Dick, C.; Neuman, A.; A1-Sadir, J.; Chua, K. G.; Carroll, J.; Feldman, T.; Borow, K. M. Contrast eehocardiography during coronary arteriography in humans: Perfusion and anatomic studies. J. Am. Coll. Cardiol. 11:59-65; 1988. Feinstein, S. B.; Cheirif, J.; ten Cate, F. J.; Silverman, P. R.; Heidenreich, P. A.; Dick, C.; Desir, R. M.; Armstrong, W. F.; Quinones, M. A.; Shah, P. M. Safety and efficacy of a new transpulmonary ultrasound contrast agent: Initial multicenter clinical results. J. Am. Coil. Cardiol. 16:316-324; 1990. Gazanhes, C.; Arzelies, P.; Leandre, J. Propagation acoustique dans un miliue diphasique eau-bulles d'air. Application ~tla earacter-

Volume 19, Number 8, 1993 isation d'un miliue diphasique par voie acoustique. Acustica 55:113-122; 1984. Glantz, S. A. Primer ofbiostatistics. New York: McGraw-Hill Book Company; 1981. Goldman, M.; Mindich, B. Intraoperative cardioplegia contrast echocardiography for assessing of myocardial perfusion during heart surgery. J. Am. Coll. Cardiol. 4:1029-1043; 1987. Gonzalez-Fernandez, J. M. Theory of the measurement of the dispersion of an indicator in indicator dilution studies. Circ. Res. 10:409-428; 1962. Grill, H. P.; Brinker, J. A.; Taube, J. C.; Walford, J. D.; Midei, M. G.; Flaherty, J. T.; Weiss, J. L. Contrast echocardiographic mapping of collateralized myocardium in humans before and after coronary angioplasty. J. Am. Coll. Cardiol. 16:1594-1600; 1990. Heymann, M.; Payne, B. D.; Hoffman, J. I.; Rudolph, A. M. Blood flow measurements with radionuclide-labeled particles. Prog. Cardiovasc. Dis. 20:55-78; 1977. Jaffe, C. C.; Harris, D. J. Physical factors influencing numerical echo-amplitude data extracted from B-scan ultrasonic images. J. Clin. Ultrasound 8:327-333; 1980. Kaul, S.; Pandian, N. G.; Okada, R. D.; Pohost, G. M.; Weyman, A. E. Contrast echocardiography in acute myocardial ischemia: In vivo determination of total left ventricular area at risk. J. Am. Coll. Cardiol. 4:1274-1291 ; 1984. Kaul, S.; Glasheen, W.; Ruddy, T. D.; Pandian, N. G.; Weyman, A. E.; Okada, R. D. The importance of defining left ventricular area at risk in vivo during acute myocardial infarction: An experimental evaluation with myocardial contrast two-dimensional echocardiography. Circulation 75:1249-1260; 1987a. Kaul, S.; Pandian, N. G.; Guerrero, J. C.; Gillam, L. D.; Okada, R. D.; Weyman, A. E. Effects of selectively altering the collateral pressure on regional perfusion and function in the coronary bed in the dog. Circ. Res. 61:77-85; 1987b. Kaul, S.; Kelly, P.; Oiiner, J. D.; Glasheen, W. P.; Keller, M. W.; Watson, D. D. Assessment of regional myocardial blood flow with myocardial contrast two-dimensionalechocardiography. J. Am. Coll. Cardiol. 13:468-482; 1989. Kaul, S. Quantitation of myocardial perfusion with contrast echocardiography. Am. J. Cardiac Imaging 5:200-216; 1991. Kaul, S.; Glasheen, W. P.; Oliner, J. D.; Kelly, P.; Gascho, J. A. Relation between anterograde blood flow through a coronary artery and the size of the perfusion bed it supplies: Experimental and clinical implications. J. Am. Coll. Cardiol. 17:1403-1413; 1991. Keller, M. W.; Feinstein, S. B.; Briller, R. A.; Powsner, S. M. Automated production and analysis of echo contrast agents. J. Ultrasound Med. 5:493-498; 1986. Keller, M. W.; Feinstein, S. B.; Watson, D. D. Successful left ventricular opacification following peripheral venous injection ofsonieated contrast agent: An experimental evaluation. Am. Heart J. 114:570-575; 1987. Keller, M. W.; Giasheen, W.; Smucker, M. L.; Burwell, L. R.; Watsonn, D. D.; Kaul, S. Myocardial contrast eehocardiography in humans. II. Assessment of coronary blood flow reserve. J. Am. Coll. Cardioi. 12:925-934; 1988a. Keller, M. W.; Glasheen, W.; Teja, K.; Gear, A.; Kaul, S. Myocardial contrast eehoeardiography without significant hemodynamic effects or reactive hyperemia: A major advantage in the imaging of regional myocardial perfusion. J. Am. Coll. Cardiol. 12:1039-1047; 1988b. Keller, M. W.; Segal, S. S.; Kaul, S.; Duling, B. The behavior of sonieated albumin microbubbles within the microcirculation: A basis for their use during myocardial contrast echocardiography. Circ. Res. 65:458-467; 1989. Keller, M. W.; Spotnitz, W. D.; Matthew, T. L.; Glasheen, W. P.; Watson, D. D.; Kaul, S. Intraoperative assessment of regional myocardial peffusion using quantitative myocardial contrast echoeardiography: An experimental evaluation. J. Am. Coll. Cardiol. 16:1267-1279; 1990. Kemper, A. J.; Force, T.; Kloner, R. A.; Gilfoil, M.; Perkins, L.; Hale, S.; Alker, K.; Parisi, A. F. Contrast echocardiographic

Myocardial perfusion measurement by contrast echocardiography • V. MoR-Av] et al. estimation of regional myocardial blood flow after acute coronary occlusion. Circulation 72:1115-1124; 1985. Kerber, R. E.; Marcus, M. L.; Ehrhardt, J.; Wilson, R.; Abboud, F. M. Correlation between echocardiographically demonstrated segmental dyskinesis and myocardial perfusion. Circulation 52:1097-1104; 1975. LaFollette, P. S.; Ziskin, M. C. Geometric and intensity distortion in echography. Ultrasound Med. Biol. 12:953-963; 1986. Lang, R. M.; Feinstein, S. B.; Feldman, T.; Neuman, A.; Chua, K. G.; Borow, K. M. Contrast echocardiography for evaluation of myocardial peffusion: Effects of coronary angioplasty. J. Am. Coll. Cardiol. 8:232-235; 1986. Marcus, M. L.; Kerber, R. E.; Ehrhardt, J.; Abboud, F. M. Three dimensional geometry of acutely ischemic myocardium. Circulation 52:254-263; 1975. Marcus, M. L.; Wilson, R. F.; White, C. W. Methods for measurement of myocardial blood flow in patients: A critical review. Circulation 76:245; 1987. Martin, R. P. Myocardial contrast echocardiography: A light in the heart of darkness. J. Am. Coll. Cardiol. 13:857-859; 1989. Meerbaum, S. Can contrast echocardiography elucidate myocardial physiology?J. Am. Coll. Cardiol. 17:1414-1415; 1991. Meier, P.; Zierler, K. L. On the theory of the indicator dilution method for measurement of blood flow and volume. J. Appl. Physiol. 6:731-744; 1954. Melton, H. E.; Skorton, D. J. Rational gain compensation for attenuation in cardiac ultrasonography. Ultrason. Imaging 5:214228; 1983. Meltzer, R. S.; Tickner, E. G.; Sahines, T. P.; Popp, R. L. The source of ultrasound contrast effect. J. Clin. Ultrasound 8:121127; 1980. Mor-Avi, V.; Akselrod, S.; David, D.; Keselbrener, L. Digital pathdependent recompensation of contrast-enhanced echocardiographic images. Ultrasound Med. Biol. 18:831-842; 1992. Mor-Avi, V.; Akselrod, S.; David, D.; Keselbrener, L.; Bitton, Y. Myocardial transit time of the echocardiographic contrast media. Ultrasound Meal. Biol. 19:635-648, 1993. Morse, P. M.; Ingard, K. U. Theoretical acoustics. New York: McGraw-Hill Book Co.; 1968:400-441. Nishi, R. Y. The scattering and absorption of sound waves by a gas bubble in a viscous liquid. Acustica 33:65-74; 1975. O'Donnell, M. Quantitative volume backscatter imaging. IEEE Trans. Sonics Ultrasonics 30:26-36; 1983. Pincu, M.; Schwartz, G.; Corday, S. R.; Fujibayashi, Y.; Meerbaum, S. Attenuation correction in echocardiography. Ultrason. Imaging 8:86-106; 1986. Powsner, S. M.; Keller, M. W.; Saniie, J.; Feinstein, S. B. Quantitation of echo-contrast effects. Am. J. Physiol. Imaging 1:124128; 1986. Pye, S. D.; Wild, S. R.; McDicken, W. N. Adaptive time gain compensation for ultrasonic imaging. Ultrasound Med. Biol. 18:205-212; 1992. Rayleigh, J. W. S. The theory of sound, 2nd ed. New York: Dover Publishers; 1945. Reisner, S. A.; Ong, L. O.; Lichtenberg, G. S.; Amico, A. F.; Shapiro, J. R.; Allen, M. N.; Meltzer, R. S. Myocardial perfusion imaging by contrast echocardiography with use ofintracoronary sonicated albumin in humans. J. Am. Coll. Cardiol. 14:660665; 1989. Rovai, D. R.; Nissen, S. E.; Elion, J.; Smith, M.; L'Abbate, A.; Kwan, O. L.; DeMaria, A. N. Contrast echo washout curves from the left ventricle: Application of basic principles of indicator dilution theory and calculation of ejection fraction. J. Am. Coll. Cardiol. 10:125-134; 1987. Shapiro, J. R.; Reisner, S. A.; Meltzer, R. S. Prospects of transpulmonary contrast echocardiography. J. Am. Coll. Cardiol. 13:1629-1630; 1989. Shapiro, J. R.; Reisner, S. A.; Amieo, A. F.; Kelly, P. F.; Meltzer, R. S. Reproducibility of quantitative myocardial contrast echocardiography. J. Am. Coll. Cardiol. 15:602-609; 1990a. Shapiro, J. R.; Reisner, S. A.; Lichtenberg, G. S.; Meltzer, R. S. Intravenous contrast echocardiography with use of sonicated

633

albumin in humans: Systolic disappearance of left ventricular contrast after transpulmonary transmission. J. Am. Coll. Cardiol. 16:1603-1607; 1990b. Smith, M. D.; Elion, J. L.; McClure, R. R.; Kwan, O. L.; DeMaria, A. N.; Evans, J.; Fritzsch, T. H. Left heart opacification with peripheral venous injection of a new saccharide echo contrast agent in dogs. J. Am. Coll. Cardiol. 13:1622-1628; 1989. Tei, C.; Sakamaki, T.; Shah, P. M.; Meerbaum, S.; Shimoura, K.; Kondo, S.; Corday, E. Myocardial contrast echocardiography: A reproducible technique of myocardial opacification for identifying regional perfusion defects. Circulation 67:585-594; 1983. Trautman, E. D.; Newbower, R. S. The development of indicator dilution techniques. IEEE Trans. Biomed. Eng. 31:800-807; 1984. Vanderberg, B. F.; Feinstein, S. B.; Kieso, R. A.; Hunt, M.; Kerber, R. E. Myocardial risk area and peak gray level measurement by contrast echocardiography: Effect of microbubble size and concentration, injection rate, and coronary vasodilation. Am. Heart J. 115:733-739; 1988. Vanderberg, B. F.; Kieso, R.; Fox-Eastham, K.; Chilian, W.; Kerber, R. E. Quantitation of myocardial perfusion by contrast echocardiography: Analysis of contrast gray level appearance variables and intracyclic variability. J. Am. Coll. Cardiol. 13:200-206; 1989. Villanueva, F. S.; Glasheen, W. P.; Sklenar, J.; Jayaweera, A. R.; Kaul, S. Successful and reproducible myocardial opacification during two-dimensional echocardiography from right heart injection of contrast. Circulation 85:1557-1564; 1992. Widimsky, P.; Cornel, J. H.; Ten Care, F. J. Evaluation of collateral blood flowby myocardial contrast enhanced echocardiography. Br. Heart J. 59:20-22; 1988. Zierler, K. L. A simplified explanation of the indicator dilution for measurement of fluid flow and volume and other distributive phenomena. In: The John Hopkins Proceedings; 1958:199217. Zierler, K. L Theoretical basis of indicator dilution methods for measuring flow and volume. Circ. Res. 10:393-407; 1962. NOMENCLATURE

V

~

f= (t) =

cp(t) =

M~(t) =

J=

j=

v o l u m e o f a s a m p l e o f m y o c a r d i a l tissue; i n t r a v a s c u l a r fraction o f v o l u m e V; m e a n transit time of microbubbles through the tissue sample; c o n c e n t r a t i o n o f b u b b l e s i n the capillaries i n tissue sample; concentration of bubbles in a homogen e o u s b l o o d pool u p s t r e a m the tissue sample, such as the left v e n t r i c u l a r cavity or the aortic root; t r a c e r - p r o d u c e d u l t r a s o n i c reflection intensity from the tissue sample; t r a c e r - p r o d u c e d reflection i n t e n s i t y from the b l o o d pool; total n u m b e r o f b u b b l e s i n v o l u m e V o f t h e tissue sample; total n u m b e r o f b u b b l e s i n v o l u m e V o f b l o o d i n the pool; total b l o o d flow t h r o u g h the tissue s a m p l e (in m L / m i n ) ; p e r f u s i o n i n the s a m p l e tissue (in m L / m i n / lO0 g); arbitrary c o n s t a n t s .