Myocardial border tracking in M-mode echocardiograms using joint process estimation

Abstract

A sequential approach to the detection of myocardial borders in M-mode echocardiograms is introduced. Initial border estimates are obtained from a cross-correlation detector. They are then improved by an adaptive lattice-form joint process predictor. Alternatively, a physiological constraint is used to improve the detection of the endocardium during systole. A least squares algorithm is proposed to update recursively the correlation templates in order to track their temporal variations.<>