Fuzzy neural network applications in medicine

A fuzzy neural network based approach to 3-D heart motion understanding is proposed. Knowledge from cardiologists is used to specify different classes of motion and classification rules. The objective of this approach is to find the decisions for all possible classes of motion in the form of possibilities. The neural networks in the motion understanding system are independent of each other in operation and are cascaded to form a tree structure where each terminal node represents one possible class and each nonterminal node composed of a neural network represents one classification rule. In each network, the propagation rule is described by a fuzzy function and a supervised learning method is employed to train the parameters representing the type of fuzzy logic operation among inputs and the weights between each input and output neurons. Decision values can be computed by applying fuzzy reasoning technique to the outputs of the networks. Experiments on real data have been conducted to corroborate the proposed techniques