SUPERVISED INTERPRETATION OF ECHOCARDIOGRAMS WITH A PSYCHOLOGICAL MODEL OF EXPERT SUPERVISION

We have developed a collaborative scheme that facilitates active human supervision of the binary segmentation of an echocardiogram. The sche me complements the reliability of a human expert with the precision of segmentation algorithms. In the developed system, an expert user comp ares the computer generated segmentation with the original image in a user friendly graphics environment, and interactively indicates the in correctly classified regions either by pointing or by circling. The pr ecise boundaries of the indicated regions are computed by studying ori ginal image properties at that region, and a human visual attention di stribution map obtained from the published psychological and psychophy sical research. We use the developed system to extract contours of hea rt chambers from a sequence of two dimensional echocardiograms. We are currently extending this method to incorporate a richer set of inputs from the human supervisor, to facilitate multi-classification of imag e regions depending on their functionality. We are integrating into ou r system the knowledge related constraints that cardiologists use, to improve the capabilities of our existing system. This extension involv es developing a psychological model of expert reasoning, functional an d relational models of typical views in echocardiograms, and correspon ding interface modifications to map the suggested actions to image pro cessing algorithms.