[Three-dimensional image processing for analysis of ventricular wall motion].

Abstract

Three-dimensional data processing was developed for single photon computed tomographic imaging of the multigated blood pool with Tc-99m labelled red blood cells. A gamma camera (ZLC-7500; Siemens) was rotated from the left posterior oblique (LPO 45 degrees) to the right anterior oblique (RAO 45 degrees) projections at intervals of 180 degrees/32. Data were collected at each part for one to two minutes, and the number of gates per beat was 12 to 20, depending on the patient. Three-dimensional transaxial images of the heart were reconstructed by the filtered back projection method without attenuation correction. A Shepp & Logan filter was used for the convolution filter. Short-axial images were reconstructed from the transaxial images. Using the equicount level method, the contours of both ventricles at each short-axial cross-section image were determined automatically with 50% levels of the maximum counts. Each contour line was smoothed to the fifth order of Fourier function after conversion from (x-y) coordinates to (r-theta) polar coordinates with respect to the center of the area. Each contour line, converted again into (x-y) coordinates, was drawn successively on a color CRT at constant intervals and inclinations in order to display the ventricles three-dimensionally. It is possible to display simultaneously on a CRT the right and left ventricles or two types of cardiac disease. It is also possible to depict superimposed images of end-diastolic and end-systolic ventricles. Three cases including a normal control, one with old inferior infarction and one with hypertrophic cardiomyopathy (HCM) were examined.(ABSTRACT TRUNCATED AT 250 WORDS)