Volume 113 Number 6 Doppler predictions of pulmonary hemodynamics day variability in normal subjects. Am J Cardiol 1984; 54:1092-1098. 43. Gardin JM, Iseri LT, Elkayam U, Tobis J, Childs W, Burn CS, Henry WL. Evaluation of dilated cardiomyopathy by pulsed Doppler echocardiography. AM HEART J 1984; 106:1057-1065. 44. Rich S, D’Alonzo GE, Dantzker DR, Levy PS. Magnitude and implications of spontaneous hemodynamic variability in pri- mary pulmonary hypertension. Am J Cardiol 1985,55:159- 163. Limitations of digital subtraction contrast echocardiography in enhancing left ventricular endocardial definition The purpose of this study was to evaluate whether the digital subtraction technique, applied to contrast echocardiography of the left ventricle (LV), might improve endocardial edge identification by two-dimensional echocardiography. Injections of the polysaccharide agent SHU-454 were made into the LV of five closed-chest dogs. Data were obtained at different levels of ejection fraction (EF) induced by pharmacologic or mechanical interventions and were documented by left ventriculography (VGRAM) In the right anterior oblique projection. Contrast echocardtography was recorded in the apical four-chamber view. The echocardiographk Images were digitized off-line into a 256 X 256 pixel matrix with 256 gray levds/plxel. Two end-diastolic frames prior to contrast appearance were averaged to obtain a mask that was subtracted from end-diastolic contrast frames corresponding to the two beats of peak intensity. The same procedure was repeated for the systolic frames. LV edges from echocardiographic images prior to contrast appearance, from digitally subtracted echo-contrast images, and from VGRAM were traced on two occasions by two different observers. LV volumes were calculated by rlnglsplane Simpson’s rule and EF was derived by the classical equation. The lntra and interobserver reproducibility in the measurement of EF was excellent for VGRAM (r = 0.95 and 0.94, respectively), It was good for two-dimensional echocardiography (I = 0.87 and 0.73), and was fair for contrast-echo (f = 0.79 and 0.68). Estimates of EF by two-dimensional echocardiography correlated with those of VGRAM (I = 0.82) slightly better than with those of contrast-echo (r = 0.76). In conclusion, in thls experimental study, the application of dtgttal subtraction to contrast echocardlography of the LV did not help to improve LV edge identification by two-dimensional echocardiography, as evaluated by the reproducibility and accuracy in EF calculation. (AM HEART J 1987;113:1437.) Daniele Rovai, M.D.,** Steven E. Nissen, M.D.,* Jonathan L. Elion, M.D.,* Alessandro Distante, M.D.,** and Anthony N. DeMaria, M.D.* Lexington, KY., and Pisa, Italy Accurate identification of the endocardial border of accurate determination of a variety of measure- the left ventricle is of critical importance for the ments by echocardiography, including wall thick- ness, ventricular volumes, and ejection fraction. From the *Division of Cardiovascular Medicine, University of Kentucky However, detailed delineation of the endocardial College of Medicine and Veterans Administration Medical Center; and surface by echocardiography is often difficukl **C.N.R. Clinical Physiology Institute, University of Pisa. Dropout of the endocardial signal frequently occurs, Supported by Veterans Administration Merit Review Grant No. 596- particularly in imaging planes in which the ultrason- 76-9148-001; and by a CNR-NATO fellowship (Dr. Rovai). ic beam is relatively parallel to the left ventricular Received for publication Oct. 10, 1986; accepted Nov. 20, 1986. Reprint requests: Anthony N. DeMaria, M.D., Division of Cardiovascular walls or when there is impediment to the transmis- Medicine, University of Kentucky Medical Center, 800 Rose St., Lexing- sion of ultrasound imposed by thoracic anatomy. ton, KY 40536. Conversely, ultrasonic noise may obscure the border 1437