We developed a technique to measure regional two-dimensional deformations in the myocardium. Three piezoelectric crystals were implanted in a triangular array in the left ventricular anterior midwall in six anesthetized dogs. Each crystal was used in a dual function, to both transmit and receive ultrasonic signals from the other two crystals. In this manner, the three segment lengths of the crystal triangle throughout the cardiac cycle were simultaneously recorded. The orientation of the crystal triangle with reference to the left ventricular long and minor axes was determined. The orientation and three segment lengths of the crystal triangle were used to calculate the circumferential strain E11, the longitudinal strain E22, the in-plane shear strain E12, and the mutually perpendicular principal strains E1 and E2. Also, the orientation of the first principal direction or the in-plane angle was determined, which was defined as the angle between the first principal direction (E1) and the circumferential direction (0 degree). This information fully describes the regional two-dimensional myocardial deformations. This technique was applied to measure regional myocardial deformations at three different left ventricular end-diastolic pressures (LVEDP) of 2 +/- 1 (mean +/- SD), 8 +/- 1, and 17 +/- 2 mm Hg. The first principal direction at end-systole was oriented away from the circumferential direction at low LVEDP (-43 +/- 21 degrees) but became progressively closer in each animal to the circumferential direction as LVEDP increased to mid (-26 +/- 18 degrees) and high (-14 +/- 13 degrees) levels. The end-systolic ratio E11/E1 was 0.6 +/- 0.2 at low LVEDP, but increased toward unity in each animal to 0.9 +/- 0.1 at mid and high LVEDP. Thus, at low LVEDP, the greatest systolic deformation occurred in a direction different from the circumferential orientation. Therefore, circumferential strain measurements (E11) significantly underestimated the greatest systolic deformation (E1). However, as LVEDP increased, the first principal direction rotated closer toward the circumferential orientation, and circumferential strain measurements adequately estimated the greatest systolic deformation. Nevertheless, the presence of significant amounts of shortening along either the longitudinal (E22) or the second principal direction (E2) in the midwall necessitated the use of the two-dimensional method. The change in end-diastolic configuration as LVEDP increased from 1 +/- 1 to 16 +/- 1 mm Hg was also examined. Unlike the end-systolic data, the end-diastolic first principal direction did not deviate significantly from the circumferential direction at any LVEDP.(ABSTRACT TRUNCATED AT 400 WORDS)