Shape based classification of left ventricles by spectral analysis of their geometrical cardiograms

摘要

The recently introduced geometrical cardiogram (GCG) is an anatomically aligned helical vector, normalized to the size of the heart, that characterizes the three-dimensional instantaneous shape of the left ventricle. It is shown that the spectrum obtained by applying a Fourier sine series expansion to the end-systolic (ES) GCG can be utilized to classify normal, hypertension/hypertrophic (HT), and infarcted (MI) hearts. No human intervention is required once the endocardial contours are traced. The demonstrated high discriminatory potential of the mod A/sub 1/ mod , and GR/sub 8/ parameters implies that a fully automated diagnostic procedure is potentially obtainable.