LESION SIZE QUANTIFICATION IN SPECT USING AN ARTIFICIAL NEURAL-NETWORK CLASSIFICATION APPROACH

An artificial neural network (ANN) has been developed to determine the size of lesions detected in single photon emission computed tomograph ic images. The network is the Learning Vector Quantizer and is trained to perform size quantification based on image neighborhoods extracted around the lesions. The ANN is compared to the optimal, Bayesian algo rithm developed to perform the same task using the unreconstructed, pr ojection data. The performance of the neural network is evaluated at t wo different noise levels. The Bayesian algorithm provides the upper b ound for size quantification performance against which the ANN is comp ared. In the ideal case where the Bayesian algorithm has explicit know ledge of the underlying distributions, its performance is superior to that of the neural network. However, in the more realistic case where the distributions need to be estimated from the same learning sample t he ANN was trained on, the two algorithms have comparable performances . (C) 1995 Academic Press, Inc.