A NEURAL NETWORK THAT LEARNS TO INTERPRET MYOCARDIAL PLANAR THALLIUM SCINTIGRAMS

The planar thallium-201 (Tl-201) myocardial perfusion scintigram is a widely used diagnostic technique for detecting and estimating the risk of coronary artery disease. Interpretation is currently based on visu al scoring of myocardial defects combined with image quantitation and is known to have a significant subjective component. Neural networks l earned to interpret thallium scintigrams as determined by both individ ual and multiple (consensus) expert ratings. Four different types of n etworks were explored: single-layer, two-layer backpropagation (BP), B P with weight smoothing, and two-layer radial basis function (RBF). Th e RBF network was found to yield the best performance (94.8% generaliz ation by region) and compares favorably with human experts. We conclud e that this network is a valuable clinical tool that can be used as a reference ''diagnostic support system'' to help reduce inter- and intr aobserver variability. This system is now being further developed to i nclude other variables that are expected to improve the final clinical diagnosis.