Ultrasonic backscatter system for automated on-line endocardial boundary detection: evaluation by ultrafast computed tomography

Findings of automatic border detection in subjects with left ventricular diastolic dysfunction by Doppler echocardiography

Left ventricular automatic border detection (ABD) patterns were defined in the parasternal short-axis and the apical four-chamber views and were compared with pulsed-wave diastolic Doppler flow-velocity patterns of the mitral valve in 49 subjects (aged 39 to 87 years), 10 selected normal individuals, and 39 consecutive patients with high-quality echocardiographic Doppler studies and relaxation abnormalities (Doppler peak early diastolic velocity/peak late velocity ratio < 1). Both short-axis and apical four-chamber views were useful in the assessment of diastolic function by ABD. However, in subjects with high-quality two-dimensional echocardiographic Doppler studies, ABD was technically more feasible in the apical four-chamber view (97%; 38/39 subjects) than in the short-axis view (64%; 25/39 subjects) and correlated better with Doppler parameters. Compared with normal subjects, patients with abnormal Doppler relaxation patterns showed significant differences in diastolic filling indexes obtained by the ABD technique. The ratio of peak rapid filling rate/peak atrial filling rate (PRFR/PAFR) obtained from the dA/dt waveform in the apical four-chamber view had the highest correlation with Doppler indexes (r = 0.79). A PRFR/PAFR ratio of 1.5 best discriminated between normal individuals and subjects with relaxation abnormalities, with high sensitivity and specificity (95% and 100%, respectively). The PRFR/PAFR ratio obtained from the dA/dt waveform seemed to be a simple and useful method to distinguish between normal and abnormal left ventricular diastolic filling, as defined by Doppler echocardiography.

Measurement of left ventricular ejection fraction by acoustic quantification and comparison with radionuclide angiography

Left ventricular (LV) ejection fraction (EF) is an important measure of systolic function, with radionuclide angiography being the accepted standard for its determination. Echocardiography is ideal for repeated measurements of EF, but most methods are either subject to error in the presence of regional wall abnormalities or require cumbersome off-line analysis. Acoustic quantification is a recently introduced method that allows for the continuous on-line display of LV cavity dimensions, but the on-line algorithm for the measurement of EF has not been validated against an independent standard in the clinical setting. This study attempted to validate acoustic quantification in the determination of EF by comparison with off-line echocardiographic analysis and radionuclide angiography in 54 patients referred for this latter procedure. Acoustic quantification correlated well with off-line analysis in both the apical 4-chamber (r = 0.89, n = 43) and 2-chamber (r = 0.86, n = 26) views. Similarly, it also correlated well with radionuclide angiography in the 4-chamber (r = 0.81, n = 44) and 2-chamber (r = 0.83, n = 26) views. The correlation between the 2 methods was further improved when only the last 30 patients were assessed (r = 0.91, n = 25 for 4-chamber views; r = 0.86, n = 16 for 2-chamber views). The correlation was worse in patients with regional asynergy (r = 0.69, n = 17 for 4-chamber views; r = 0.76, n = 10 for 2-chamber views). Moreover, acoustic quantification tended to underestimate EF when compared with radionuclide angiography.(ABSTRACT TRUNCATED AT 250 WORDS)

Determination of cardiac output using acoustic quantification in critically ill patients

The reliability of automated acoustic quantification (AQ) in yielding real-time left ventricular (LV) area, volume, and ejection fraction has been validated. The purpose of this study was to explore the potential of AQ in providing accurate immediate determination of cardiac output in critically ill patients. A total of 48 patients were studied. One group consisted of 37 critically ill patients with indwelling Swan-Ganz catheters. In these patients, cardiac output by AQ, manual tracing of end-diastolic and end-systolic frames of 2-dimensional images, and thermodilution were measured. AQ was also compared with Doppler calculation of flow through the left and right ventricles in 11 additional patients. Adequate data for calculation of cardiac output with AQ were obtained in 78% of critically ill patients. There was an excellent correlation between AQ and off-line manual analysis for LV volumes (r = 0.94 and 0.91 for end-diastole and end-systole), ejection fraction (r = 0.85), and cardiac output (r = 0.93). AQ also correlated well with Doppler analysis (r = 0.97) and thermodilution technique (r = 0.95) in the determination of cardiac output. However, AQ slightly underestimated thermodilution measurements, with rather wide limits of agreement (-0.3 +/- 1.1 liter/min). There was a similar underestimation of cardiac output with manual analysis when compared with thermodilution. Given the absence of significant differences between AQ and manual analysis, this observation suggests that the bias is related to the echocardiographic determination of stroke volume, and not to errors from the automated border detection. It is concluded that AQ, besides providing information on LV volumes and ejection fraction, also can yield rapid measurements of cardiac output in most patients who are acutely ill.

Measurement of regional elastic properties of the human aorta. A new application of transesophageal echocardiography with automated border detection and calibrated subclavian pulse tracings

BACKGROUND

Evaluation of regional aortic elastic properties in humans has been hampered by the need for invasive techniques to access instantaneous aortic pressure, wall thickness, and cross-sectional area or diameter. In this study, a new noninvasive method is presented for quantification of regional aortic elastic properties.

METHODS AND RESULTS

Twenty-five patients were studied during transesophageal echocardiographic procedures. Measurements of instantaneous aortic cross-sectional area were obtained with an automated border detection algorithm applied to short-axis transesophageal two-dimensional echocardiographic images of the proximal descending thoracic aorta. Instantaneous aortic wall thickness was derived from combined two-dimensional targeted M-mode end-diastolic wall thickness and instantaneous aortic area measurements. Instantaneous aortic pressures were estimated from calibrated subclavian pulse tracings recorded simultaneously. Data were digitized to generate aortic area-pressure loops. Regional aortic mechanical properties were quantified in terms of compliance per unit length (C is the slope of the area-pressure regression), aortic midwall radius (Rm), and incremental elastic modulus of the aortic wall (Einc). To assess the independent effect of age, Rm and Einc values were compared at a common level of aortic midwall stress (0.666 x 10(6) dynes/cm2). Mean values (+/- SD) for C, Rm, and Einc were 0.01 +/- 0.004 cm2/mm Hg, 1.14 +/- 0.17 cm, and 7.059 +/- 4.091 x 10(6) dynes/cm2, respectively. An inverse linear correlation was found between aortic compliance per unit length and age (r = -.68, P < .0007). Incremental elastic modulus was related to age (r = +.80, P < .00003) in a nonlinear fashion such that it increased sharply after the age of 60 years. Finally, midwall radius was less tightly correlated with age (r = +.45, P < .05). Values for C, Rm, and Einc as well as the age dependency of these properties are similar to those reported previously when invasive techniques were used.

CONCLUSIONS

This methodology constitutes a new tool to improve the clinical evaluation of regional aortic elastic properties in multiple disease states.

Two-dimensional echocardiographic automatic boundary detection for evaluation of left ventricular function in unselected adult patients

Recent modifications in two-dimensional echocardiographic imaging include automatic boundary detection (ABD) that permits instantaneous quantification of left ventricular (LV) function. To define its clinical application from different views in consecutive patients (n = 68), we performed both conventional two-dimensional imaging and ABD with lateral gain compensation and compared the ability of each method for visualization of myocardial segments. From each view (short-axis and apical four- and two-chamber), the LV was divided into six wall segments. The short-axis view was obtained successfully in 53 of 68 patients; of the possible 318 segments, 96% were visualized adequately by two-dimensional echocardiography and 89% by ABD. From the four-chamber view, obtained in 63 of 68 patients, visualization of 378 possible segments was 93% by two-dimensional imaging and 86% by ABD, respectively. From the two-chamber view obtained in 58 of 68 patients, the success rate was 88% and 80% (two-dimensional imaging and ABD, respectively). ABD detected LV endocardial/blood borders in all six segments of each view in 73%, 72%, and 72% of cases, concordant with adequate two-dimensional imaging. Discrepancies between two-dimensional imaging and ABD were usually in the short-axis inferior and lateral, four-chamber lateral, and two-chamber anterior segments. There were 46 patients with normal LV wall motion and 22 with wall-motion abnormalities. The average time required to perform ABD was 375 seconds (range 180 to 780 seconds).(ABSTRACT TRUNCATED AT 250 WORDS)

Utility and limitations of biplane transesophageal echocardiographic automated border detection for estimation of left ventricular stroke volume and cardiac output

Recent echocardiographic ABD algorithms can estimate LV volume on-line from a single long-axis plane. The objective of this study was to assess the capability and limitations of transesophageal ABD to estimate stroke volume and cardiac output in patients before and after coronary artery bypass surgery by correlating these data with simultaneous thermodilution measurements. ABD data were acquired on-line from the transverse-plane four-chamber view and the longitudinal-plane two-chamber view and calculated by automated area-length and Simpson's rule formulas for volume. Thirty-three studies were attempted in 18 patients. Technically adequate ABD data were available in all patients from at least one view. Twenty-two (67%) of 33 studies from the four-chamber view and 27 (82%) of 33 studies from the two-chamber view were technically adequate. Cardiac output by all ABD methods was significantly correlated with thermodilution values (r range 0.72 to 0.89; SEE range 0.48 to 0.55 L/min). The two-chamber view underestimated cardiac output slightly, by an average of 0.4 L/min, whereas the four-chamber view consistently underestimated cardiac output by an average of 1.9 L/min. The area-length and Simpson's rule algorithms produced similar results. Biplane transesophageal ABD is an alternative method for estimating cardiac output; the two-chamber view in particular has potential for on-line volume determination.

Quantitative Echocardiography, Part 2

Two-dimensional echocardiography is widely used for assessment of cardiac chamber size and function. A new method for real-time quantitative assessment of left ventricular systolic function using automatic boundary detection (ABD) has been developed based on principles of quantitative ultrasound backscatter imaging of the myocardium. The clinical accuracy of ABD for measurement of cardiac chamber areas and volumes has been validated against conventional two-dimensional echocardiography, transesophageal echocardiography, radionuclide ventriculography, and left ventriculography. Left ventricular diastolic function can also be evaluated by ABD, and the results complement those obtained indirectly by pulsed Doppler echocardiographic recordings of the diastolic mitral flow velocity. Recently, ABD has also been used to evaluate left atrial chamber dimensions and atrial function. Thus, real-time ABD is a new echocardiographic method for quantitative assessment of cardiac function that obviates the need for off-line analysis of video images.