Two-dimensional echocardiographic imaging: In vitro comparison of conventional and dynamically focused annular array transducers

Ultrasound in Med & Biol, Vol. 13, No. 10, pp. 643-650, 1987 Printed in the U.S.A.

0301-5629/87 $3.00 + .00 © 1987 Pergamon Journals Ltd.

OOriginal Contribution TWO-DIMENSIONAL ECHOCARDIOGRAPHIC IMAGING: I N VITRO COMPARISON OF CONVENTIONAL AND DYNAMICALLY FOCUSED ANNULAR ARRAY TRANSDUCERS RICCARDO PINI Division of Cardiology, New York Hospital-Cornell Medical Center, New York, NY L U I G I F E R R U C C I a n d M A U R O D I BARI Institute of Gerontology and Geriatrics, University of Florence, Florence, Italy BARBARA GREPPI a n d MARINO CEROFOLINI OTE Biomedica, Florence, Italy LEONARDO MASOTTI Department of Electronic Engineering, University of Florence, Florence, Italy

and R I C H A R D B. D E V E R E U X Department of Medicine, New York Hospital-Cornell Medical Center, New York, NY

(Received 30 December 1986; in fmalJbrm 13 April 1987) Abstract--Quantitative two-dimensional echocardiography has been adversely affected by a tendency for underestimation of cross-sectional areas of cardiac chambers, a difficulty that might be ameliorated by recent advances in imaging technology. To determine if this were so, we measured echocardiographic cross-sectional areas of 25 formalin-fixed animal left ventricular (LV) sections in vitro using conventional 13 mm and 15 mm diameter fixed-focused single element transducers, and a 15 mm diameter dynamically focused annular array transducer at 3 different distances between myocardial slice and transducer (2 cm, 6 cm and 10 cm) and compared the 2-dimensional echocardiographic areas to the corresponding anatomic cross-sectional areas of the same hearts. LV total and cavity area were measured by computer assisted planimetry of videotaped echo images, performed blinded to the transducer used, and photographed anatomic slices; LV myocardial area was derived by subtraction. Comparison of two-dimensional echocardiographic total, myocardial, and cavity areas with corresponding anatomic measurements showed excellent correlation for each transducer at all depths (r = 0.97 to 0.98 for total area; r = 0 . 9 8 t o 0 . 9 9 for cavity area; r = 0 . 9 3 t o 0 . 9 7 for myocardial area). For total and myocardial cross-sectional areas, the slope of the relation between echographic and anatomic areas did not differ significantly from unity, but for LV cavity area this was achieved only by the dynamically focused transducer. In contrast, the conventional 13 mm transducer significantly underestimated larger LV cavity areas in both the near and middle fields (slopes = 0.90 and 0.91, respectively) and the 15 mm transducer yielded slopes from 0.86 to 0.91 in all fields. The results of this study support the theoretical prediction that dynamic focusing and narrowing of beam width by annular array technology would ameliorate underestimation of cardiac chamber cross-sectional areas by two-dimensional echocardiography.

Key Word: Echocardiography. Two-dimensional echocardiography has been established as a useful noninvasive technique to measure left ventricular myocardial and cavity cross-sectional areas (Eaton e! a[., 1979; Wyatt et al., 1979; Gueret et al., 1980; Helak et al., 1981a; G o r d o n et al., 1983; Collins et al., 1984; Conetta et al., 1984; Conetta et al., 1985). Although methods with acceptable levels

of accuracy and reproducibility have been reported (Wyatt et al., 1979; Gueret et al., 1980; Helak et al., 1981a, b; Gordon et al., 1983; Wyatt et al., 1983; Collins et al., 1984; Conetta et al., 1984; Conetta et al., 1985), a systematic overestimation of the left ventricular myocardial area and underestimation of cavity area compared with actual anatomic measurements has been demonstrated by some investigators (Helak et al., 198 la; Conetta et al., 1984). Analogous underestimation of the cross-sectional area of a simulated mitral valve orifice was also encountered by

Address for correspondence: Riccardo Pini, M.D., Division of Cardiology, Box 222, New York Hospital-Cornell Medical Center, 525 East 68th Street, New York, NY 10021. 643

644

Ultrasound in Medicineand Biology

October1987, Volume 13, Number 10

Specimen collection Six animal hearts (3 pigs and 3 calves) were fixed in formalin for 24 hours, during pressure expansion (10 mmHg) of the left ventricle (Glagov et al., 1963), and subsequently hand-sectioned in the short axis plane into 3 to 5 sections ( 1.0-2.0 cm thick). In all, 25 sections were studied. The fight ventricular free wall was removed from the LV according to the chamber dissection method of Bove et al. (1966).

with a PDP 11/44 computer (Digital Equipment Corporation, Maynard, MA, USA) with: • 512 Kbyte of internal memory; • 1 Priam 15450 Winchester disk (Priam, San Jose, CA, USA) with a storage capacity of 150 Mbyte; • 7 VD1100 alpha-numeric video terminals (E.C.S. Tesak, Florence, Italy); and • 1 BitPadOne tablet (Summagraphics, Fairfield, CO, USA) with a 0.1 mm resolution. The calibration of the video system was checked with a millimetric grid. Each section was vertically suspended by two sutures on a specially designed stand placed in a glass tank filled with mineral oil. To avoid ultrasound reflections, the base of the stand was tilted at an angle of 30 ° from the horizontal. Two-dimensional echocardiographic images of each section were recorded in a short axis orientation using a commercially available sector scanner (SIM 3000, OTE Biomedica, Florence, Italy) directly connected with the image processing system described above. Each section was visualized with three different 3.5 mHz mechanical transducers (13 mm and 15 mm diameter fixed-focused single element, and 15 mm diameter dynamically focused annular array) at three different distances from the transducer to the nearest surface of the myocardial ring (2 cm, 6 cm, and 10 cm) to explore the accuracy of imaging in near, middle, and far fields. In all, 9 echocardiographic images were taken for each section (3 transducers × 3 distances), and a total of 225 sets of measurements (9 × 25 slices). The setting was adjusted to produce as complete an image as possible with the lowest gain. Correct calibration of the echocardiographic system was verified with a tissue equivalent ultrasound phantom (Mod. 412, RMI, Middleton, WI, USA). Both the anatomic and the two-dimensional echocardiographic images were stored on the digital disk and subsequently displayed to perform the measurements interactively.

Imaging techniques Images of the superior and inferior surfaces of each anatomic section with a calibration scale were acquired by a video camera (Model C150, Nordmende, Bremen, West Germany). The video camera was connected to a VDC501A image processing system (E.C.S. Tesak, Florence, Italy) with: • a frame grabber (512 × 512 pixels, 256 gray levels); • four planes for graphic overlays; and • a BARCO CDCT 2/38 high resolution color monitor (Cobar Barco Electronic, Belgium). This image processing system was interfaced

Ultrasound beam characterization The lateral resolution of the 3 different transducers (13 mm and 15 mm diameter fixed-focused single element, and 15 mm diameter dynamic focused annular array) was assessed using a truncated cone of ABS with a superior surface of 1.5 mm in diameter as target. All of the transducers were connected to the same echocardiograph (SIM 3000, OTE Biomedica Ellettronica, Firenze, Italy); the gain was set to maximum and both the power and reject to minimum while the time gain control was not modified during the acquisition of beam profiles. The target was immersed in a glass tank filled with mineral

Martin et al. (1979) under some recording conditions. This discrepancy between anatomic and echocardiographic measurements seems to be correlated with the limited lateral resolution of the conventional echocardiographs, a problem that is exacerbated when recordings are made at high gain settings. It has been recognized for some years, based on theoretical principles, that use of a "virtual ring" (Katakura et al., 1979) or annular transducer might allow the use of narrower ultrasound beams with maintenance of acceptable collimation at depths into the chest encountered in adult echocardiographic examinations. Recent improvements in technology have led to the development of echocardiographic transducers in which emitting and receiving crystals are in the form of concentric rings. Because lateral resolution is improved with an annular transducer, particularly if it is dynamically focused, more reliable measurements of the left ventricle might be expected to result from use of such a transducer. The present study was undertaken, accordingly to assess the accuracy of two-dimensional echocardiographic measurements of left ventricular cavity and myocardial crosssectional areas obtained by imaging with a single echograph in combination with a dynamic annular array mechanical transducer and two conventional single element transducers. METHODS

Two-dimensional echocardiographicimaging • R. PINI et al.

645 .15 m m DF

oil and m o v e d perpendicularly to the ultrasonic beam; the returning signal was recorded on paper using an analog plot. T h e u l t r a s o n i c b e a m was scanned at depths of 4, 6, 8, 10, 12 and 14 cm from the transducers. T h e final beam shape was reconstructed using the values o f lateral resolution obtained at - 6 dB.

E

13mm

8

v

Measurements of left ventricular total, cavity, and myocardial cross-sectional areas The left ventricular e n d o c a r d i u m and epicardium of both anatomic and two-dimensional echocardiographic images were traced directly on the monitor using the tablet. The algorithm for area calculation was validated with irregular test images of known area. The video images were outlined with the corresponding anatomic specimen as a reference; the mean of the superior and inferior surfaces was used to represent the area of each section. To be sure that the observers were blinded regarding the transducer used to visualize the anatomic sections, the echocardiographic images were coded and subsequently recalled from the digital disk using their code. The left ventricular cavity area excluded endocardial echoes, whereas the thickness of the finest outer line of epicardial echoes was excluded from total left ventricular area throughout the image. Both anatomic and echocardiographic left ventricular myocardial areas were determined by subtraction of left ventricular cavity area from total left ventricular area. Three independent observers (RP, LF, MDB) traced the first 8 photographic and 64 echocardiographic images 5 times each to test the intra- and interobserver variability; subsequently only 1 observer (RP) traced each image once. Statistical analysis The intra- and interobserver variations were examined with a two-way analysis of variance. Data from each echocardiographic transducer were compared to the corresponding photographic data by linear regression analysis and analysis of variance. The regression line for each examined relation was compared to the line o f identity for that variable using a standard statistical method for comparison of slopes (Zar, 1984). A p-value of <0.05 was accepted as significant. RESULTS The beam shape of the 3 different transducers is shown in Fig. 1. As expected, the annular array transducer presented a more regular beam along the entire

c--

10

12

14 I

1

I

I

I

1

I

I

4

3

2

I

0

2

3

4

Lateral resolution ( m m )

Fig. 1. Ultrasonic beam profiles of 13 mm diameter (dotted line) and 15 mm diameter (dashed line) fixed-focused single element transducers, and 15 mm diameter dynamic focused annular array transducer (solid line) from 4 cm to 14 cm from the transducer. Each transducer has a frequency of 3.5 mHz.

range from 4 to 14 cm of distance from the transducer; in particular, the dynamic focused transducer demonstrated the best lateral resolution (1.3 m m at 8 cm of depth) and a dramatically increased lateral resolution in the near field (2.3 m m at 4 cm of depth) compared to the fixed-focused transducers (6.3 m m with the 13 m m diameter transducer and 7.5 m m with the 15 m m diameter transducer). Also in the far field, the dynamic focused transducer maintained the best lateral resolution compared to the fixed-focused transducers (3.6 mm, 4.4 mm, and 5.0 mm, respectively). When three independent observers traced the first 8 photographic images 5 times each, the intraobserver variability was 3.5%, 4.3%, and 3.8%, respectively; these values represented the cumulative variation of cavity, total and myocardial cross-sectional areas for each observer. The echocardiographic intraobserver variations were 7.4%, 10.6%, and 7.2%. This degree of variability compares favorably with that reported by other investigators (Zwehl et al., 1981; Conetta et al., 1984; Conetta et al., 1985). Because neither intra- nor interobserver variations were statistically significant, only our observer traced each image over to analyze the differences between the 3 transducers.

646

U l t r a s o u n d in M e d i c i n e a n d Biology

O c t o b e r 1987, V o l u m e 13, N u m b e r 10

Table 1. Linear regression analyses for two-dimensional echocardiographic left ventricular total area (TAe) obtained using 13 mm and 15 mm conventional transducers, and the 15 mm dynamically focused (DF) transducer versus the photographic reference standard (TAa). (range = 27.5 to 72.1 cm2). Transducer 13 13 13 15 15 15 15 15 15

mm mm mm mm mm mm mm DF mm DF mm DF

Zone

r

Near Middle Far Near Middle Far Near Middle Far

0.98 0.97 0.98 0.98 0.98 0.97 0.98 0.98 0.98

Regression Equation TAe TAe TAe TAe TAe TAe TAe TAe TAe

= = = = = = = = =

1.03 0.95 1.03 1.02 1.00 1.05 1.03 1.04 1.02

X X X × X X × X ×

TAa TAa TAa TAa TAa TAa TAa TAa TAa

+ 0.45 + 3.36 0.01 - 0.39 0.43 - 2.35 0.79 - 2.45 - 1.64

SEE 2.637 2.658 2.531 2.426 2.155 2.987 2.279 2.362 2.625

r = c o r r e l a t i o n coefficient. SEE = standard error of estimate. n - 25 for e a c h r e g r e s s i o n e q u a t i o n . p < 0 . 0 0 1 f o r e a c h c o r r e l a t i o n coetficient.

Relation o f echocardiographic and anatomic areas A representative image of an anatomic left ventricular section with the corresponding two-dimensional echocardiographic images obtained with the 3 different transducers is shown in Fig. 1. The two-dimensional echocardiographic total left ventricular areas correlated well with the anatomic total left ventricular areas (Table 1) with each transducer at all depths; the slopes of the regression lines were close to unity (Fig. 2). The two-dimensional echocardiographic left ventricular cavity areas also correlated well (r = 0.98 to 0.99) with the anatomic ones (Table 2) under each condition. However, only the dynamically focused transducer showed a slope to the relation that was consistently near unity at each depth of field (Fig. 3). The 13 mm conventional transducer significantly underestimated larger left ventricular cavity areas in both near and middle zones; a slight underestimation was obtained with the 15 mm conventional transducer in all three zones (Fig. 4). The two-dimensional echocardiographic left ventricular myocardial areas did not differ significantly from the anatomic left ventricular myocardial areas with each transducer at any depth (Table 3), and slopes of the regression lines were close to unity in all fields for conventional and annular array transducers (Fig. 5). However, the relation between anatomic and echocardiographic measurements of myocardial area exhibited some instability at different depths of field when the 13 mm conventional transducer was used for imaging. DISCUSSION Many investigators have demonstrated that left ventricular cavity and myocardial areas can be mea-

sured by two-dimensional echocardiography with acceptable levels of accuracy and reproducibility (Eaton et al., 1979; Wyatt et al., 1979; Gueret et al., 1980; Helak et al., 1981a, b; Gordon et al., 1983; Wyatt et al., 1983; Collins et aL, 1984; Conetta et al., 1984; Conetta et al., 1985). However, systematic underestimation of left ventricular cavity area has been reported by previous investigators using both manual (Helak el al., 1981a) and automatic (Conetta et al., 1984) methods for endocardial border identification. Helak et al. (198 l a) suggested that one of the major causes for this underestimation could be related to the limited lateral resolution with consequent inaccuracy of target width estimation. If this hypothesis is true, improved lateral resolution over a wider range of depths than provided by conventional transducer types would reduce or eliminate the echocardiographic underestimation of left ventricular cavity area. Annular array transducers that are dynamically focused permit an improvement of lateral resolution all along the beam profile in each direction of planes perpendicular to the acoustic axis (Feigenbaum, 1986). This particular shape of the beam allows echocardiographic images in which artifactual overlap of the represented structures due to lateral resolution errors is not produced in any of the scanned fields. In the present study, the annular array transducer demonstrated the best lateral resolution along the entire beam profile also with a relatively small diameter of the external element (15 mm), thus suggesting it is possible to use this transducer for echocardiographic examination without the problem related to use of larger size transducers in many patients with relatively narrow intercostal spaces. As was expected

Two-dimensional echocardiographicimaging• R. P1NIel al.

647

Fig. 2. Representative calibrated photograph (A) of a short-axis section of the left ventricle and two-dimensional echocardiographic images of the same section obtained with 13 mm (B) or 15 mm (C) diameter conventional transducer and witb the 15 mm diameter (D) dynamically focused annular array transducer. The short arrow points, to the smooth epicardial surface in (A); in (B) this is displayed as a thick band of confluent interfaces; the thickness decreased progressively in (C) and (D) and the surface is displayed as a thin smooth interface in (D). The heavy long arrow points to a trabecula protruding in the left ventricular (LV) cavity in (A); in (B) the structure cannot be distinguished from the adjacent LV wall; in (C) the shape is poorly seen while in (D) the trabecula is both clearly reported from the wall and displayed with correct shape. from the beam profile of the different transducers, only with use of a dynamically focused annular array transducer did we obtain estimates of left ventricular cavity area that did not differ significantly from the actual anatomic areas at all depths of field. On the contrary, both conventional single element transducers we evaluated showed a significant underestimation of left ventricular cavity area, in accord with

previous reports (Helak et al., 1981a; Conetta et al., 1984). It is particularly noteworthy that the dynamically focused transducer, possibly because of its better beam shape, did not underestimate left ventricular cavity area significantly even though the black-white interface between cavity and endocardium was used to identify the endocardial border, a procedure reported by Conetta et al. (1984) to be less accurate

648

Ultrasound in Medicine and Biology

October1987, Volume 13, Number 10

Table 2. Linear regression analyses for two-dimensional echocardiographic left ventricular cavity area (CAe) obtained using 13 mm and 15 mm conventional transducers, and the 15 mm dynamically focused (DF) transducer versus the photographic reference standard (CAa). (range = 1.3 to 19.6 cm2). Transducer 13 13 13 15 15 15 15 15 15

mm mm mm mm mm mm mm DF mm DF mm DF

Zone

r

Near Middle Far Near Middle Far Near Middle Far

0.98 0.99 0.99 0.98 0.99 0.99 0.98 0.98 0.98

Regression Equation CAe CAe CAe CAe CAe CAe CAe CAe CAe

= = = = = = = = =

0.90 0.91 0.98 0.88 0.86 0.91 0.94 0.94 0.95

X x X X x x X X X

CAa CAa CAa CAa CAa CAa CAa CAa CAa

-

SEE

0.73 0.87 1.02 0.81 0.76 0.86 1.10 1.05 0.93

0.826 0.720 0.645 0.889 0.749 0.809 0.910 0.764 0.876

r = correlation coefficient. SEE = standard error of estimate. n = 25 for each regression equation. p < 0.001 for each correlation coefficient.

than the technique of locating the endocardium at the midpoint of the endocardial interface. However, we opted to use the black-white m e t h o d because it is m o r e reliable in that it has a smaller intra- and interobserver variability (Conetta et aL, 1984). With the annular array transducer, therefore, we were able to combine accurate m e a s u r e m e n t of the left ventricular cavity area with a m e t h o d of interface recognition previously shown to have high reliability. In the present series, we found no significant difference between single element a n d a n n u l a r array transducers in evaluating total left ventricular and m y o c a r d i a l cross-sectional areas. All three transducers slightly overestimated total and myocardial cross-sectional areas in both the near and far fields. However, in the middle field, the 13 m m transducer

underestimated the total and myocardial left ventricular areas, suggesting that this transducer yields less stable m e a s u r e m e n t accuracy at different depths in the chest than other transducers. These results m a y be related to difficulty in detecting the posterior epicardial interface in images obtained with the 13 m m transducer, which showed a reduced signal-to-noise ratio at that depth. In evaluating the results relative to the total left ventricular area, it should be taken into account that the sections presented a large external size relative to the 4-cm intervals we selected arbitrarily to define near, mid and far fields; therefore the two-dimensional echocardiographic images occupied at least two contiguous zones. The absence of significant overestimation of left ventricular myocardial area by imaging with either

1.2I

15 m m

1.1 UJ

I 13 m m

NS

NS

NS

NS

NS

15 m m DF

O 1.0 (/3 0.9.

0.8

Near

Middle TRANSDUCER

Far ZONES

Fig. 3. Comparison of slopes of each linear regression equation relating the anatomic left ventricular total area to the two-dimensional echocardiographic left ventricular total area with the line of identity. DF = dynamically focused annular array transducer. NS = not significant.

T w o - d i m e n s i o n a l e c h o c a r d i o g r a p h i c i m a g i n g • R. PINI et aL

649

Table 3. Linear regression analyses for two-dimensional echocardiographic left ve~tricular myocardial area (MAe) obtained using 13 mm and 15 mm conventional transducers, and the 15 mm dynamically focused (DF) transducer versus the photographic reference standard (MAa). (range = 26.2 to 55.7 cm2). Transducer 13 13 13 15 15 15 15 15 15

mm mm mm mm mm rnm mm DF nam DF mm DF

Zone

r

Near Middle Far Near Middle Far Near Middle Far

0.95 0.93 0.96 0.97 0.96 0.94 0.96 0.96 0.95

Regression Equation MAe MAe MAe MAe MAe MAe MAe MAe MAe

= = = = = = = =

1.06 0.95 1.05 1.06 1.07 1.11 1.05 1.07 1.04

x X x x X X X x X

MAa MAa MAa MAa MAa MAa MAa MAa MAa

+ + + + -

SEE

1.18 4.64 0.75 0.31 1.32 2.72 0.24 1.89 0.80

2.824 3.022 2.584 2.288 2.329 3.197 2.435 2.556 2.828

r = c o r r e l a t i o n coefficient. SEE = standard error of estimate. n = 25 f o r e a c h r e g r e s s i o n e q u a t i o n . p < 0 . 0 0 1 f o r e a c h c o r r e l a t i o n coefficient.

conventional or annular array transducers may relate to the small size of the LV cavity area relative to myocardial area in these sections, causing the enc r o a c h m e n t o f myocardial area into the cavity to have a proportionally smaller effect on myocardial than on cavity cross-sectional area. An additional factor likely to contribute to the better accuracy of measurement of myocardial area compared to previous studies (Helak et al., 1981a; Conetta et al., 1984) is that improvements have been made in other aspects o f echocardiographic technology in recent years. Even for myocardial area, however, the annular array transducer showed a more uniform slope of the regression lines relating echocardiographic to anatomic measurements at different depths of field than did the single element transducers.

SUMMARY The present study confirms in vitro the theoretical advantages of the annular array transducer for tw0-dimensional echocardiographic evaluation of left ventricular cross-sectional areas. Single element transducers of two sizes underestimated left ventricular cavity area, as previously reported by Helak et al. (198 l a) and Conetta et aL (1984), whereas estimates of left ventricular cavity area obtained with a dynamically focused annular array transducer did not differ significantly from the actual anatomic areas at all depths o f field. Moreover, with the annular array transducer we were able to obtain accurate measurement o f the left ventricular cavity area using the b l a c k - w h i t e m e t h o d that C o n e t t a et al. (1984)

iii"

,3 mm 15 mm

uJ a. O, 1.0 / 0 t)

~

/

/

15 mm DF

NS N5

.o,

NS

.005 ......-.-

Near

--...-.-.-

I

J

I

I.o2

NS

' " " "

Middle Far T R A N S D U C E R ZONES

Fig. 4. Comparison of slopes of each linear regression equation relating the anatomic left ventricular cavity area to the two-dimensional echocardiographic left ventricular cavity area with the line of identity. DF = dynamically focused annular array transducer. NS = not significant.

650

Ultrasound in Medicine and Biology

October 1987, Volume 13, Number 10

1.2

tu

1.1

[

I 13 m m

NS

....."-~.:.~ :~i iiiii :-!!!!!!!!!!! -ii i i i i

NS

NS

a. O 1.0 ,=,,I

NS

"-'!!!!!!!!!!-=

t,~

::::::::::::::::::

NS NS

NS

. . . . .

I~i::lIIlIIlIliII

~

15 mm

~

15 mm DF

NS

::::::::::::

0.9.

- ..-!!!!!!!!!!...

..-'!!!!!!!!!!!

0,8

|;'.'.:.:.:|11111111111

Near

liiI'lI II;II

Middle TRANSDUCER

Far

ZONES

Fig. 5. C o m p a r i s o n o f slopes of each linear regression e q u a t i o n relating the a n a t o m i c left v e n t r i c u l a r myocardial area to the t w o - d i m e n s i o n a l e c h o c a r d i o g r a p h i c left v e n t r i c u l a r myocardial area with the line o f identity. D F = d y n a m i c a l l y focused a n n u l a r array transducer. NS = n o t significant.

showed to have high reliability. Clinical studies are needed to confirm whether these promising results can be translated into improved accuracy and reproducibility of quantitative two-dimensional echocardiography. Acknowledgement--We would like to thank Miss Virginia Burns for her assistance in preparation of this manuscript.

REFERENCES Bove K. E., Rowlands D. T. and Scott R. C. (1966) Observations on the assessment of cardiac hypertrophy utilizing a chamber partition technique. Circulation 33, 558-568. Collins S. M., Skorton D. J., Geiser E. A., Nichols J. A., Conetta D. A., Pandian N. G. and Kerber R. E. (1984) Computer-assisted edge detection in two-dimensional echocardiography: comparison with a n a t o m i c data. Am. J. Cardiol. 53, 1380-1387. Conetta D. A., Geiser E. A., Skorton D. J., Pandian N. G., Kerber R. E. and Conti C. R. (1984) In vitro analysis of boundary identification techniques used in quantification of two-dimensional echocardiograms. Am. J. Cardiol. 53, 1374-1379. Conetta D. A., Geiser E. A., Oliver L. H., Miller A. B. and Conti C. R. (1985) Reproducibility of left ventricular area and volume measurements using a computer endocardial edge-detection algorithm in normal subjects. Am. ~L Cardiol. 56, 947-952. Eaton L. W., Maughan W. L., Shoukas A. A. and Weiss J. L. (1979) Accurate volume determination in the isolated ejecting canine left ventricle by two-dimensional echocardiography. Circulation 60, 320-326. Feigenbaum H. (1986) Echocardiography. Lea & Febiger, Philadelphia. Glagov S., Eckner F. A. O. and Lev M. (1963) Controlled pressure fixation apparatus for hearts. Arch. Path. 76, 640-646.

Gordon E. P., Schnittger 1., Fitzgerald P. J., Williams P. and Popp R. L. (1983) Reproducibility of left ventricular volumes by two-dimensional echocardiography. Z Am. Coll. Cardiol. 2, 506-513. Gueret P., Meerbaum S., Wyatt H. L., Uchiyama T., Lant T.-W. and Corday E. (1980) Two-dimensional echocardiographic quantitation of left ventricular volumes and ejection fraction. Importance of accounting for dysnergy in short-axis reconstruction models. Circulation 62, 1308-1318. Helak J. W., Plappert T., Muhammed A. and Reichek N. (1981a) Two-dimensional echographic imaging of the left ventricle: comparison of mechanical and phased array systems in vitro. Am. J. Cardiol. 48, 728-735. Helak J. W. and Reichek N. (1981b) Quantitation of human left ventricular mass and volume by two-dimensional echocardiography: in vitro anatomic validation. Circulation 63, 1398-1407. Katakura K., Kauda H., Ishikawa Y. and Suzuki T. (1977) Ultrasonic transducer with virtual ring image source. In Ultrasound in Medicine: 3B Engineering Aspects (Edited by D. N. White, and R. E. Brown), pp. 1849-1850. Plenum Press, New York. Martin R. P., Rakowski H., Kleiman J. H., Beaver W., London E. and Popp R. L. (1979) Reliability and reproducibility of twodimensional echocardiographic measurement of the stenotic mitral valve orifice area. Am. J. Cardiol. 43, 560-568. Wyatt H. L., Heng M. K., Meerbaum S., Hestenes J. D., Cobo J. M., Davidson R. M. and Corday E. (1979) Cross-sectional echocardiography. I. Analysis of mathematic models for quantifying mass of the left ventricle in dogs. Circulation 60, 1104-1113. Wyatt H. L., Haendchen R. V., Meerbaum S. and Corday E. (1983) Assessment of quantitative methods for 2-dimensional echocardiography. Am. J~ Cardiol. 52, 396-401. Zar J. H. (1984) Biostatistical Analysis, pp. 292-305. PrenticeHall, Inc., Englewood Cliffs. Zwehl W., Gueret P., Meerbaum S., Holt D. and Corday E. ( 1981 ) Quantitative two dimensional echocardiography during bicycle exercise in normal subjects. Am. ~L Cardiol. 47, 866-873.