GMDH-type neural networks and their application to the medical image recognition of the lungs

The GMDH (group method of data handling)-type neural networks and their application to the medical image recognition of the lungs are described. The GMDH-type neural networks have both characteristics of the GMDH and the conventional multilayered neural network and can automatically organize the optimum neural network architecture by using the heuristic self-organization method. In the GMDH-type neural networks, many types of neurons can be used to organize neural networks' architecture and neurons' characteristics which fit the complexity of the nonlinear system. They are automatically selected by using the error criterion defined as AIC (Akaike's information criterion). Therefore, many types of nonlinear systems can be automatically modeled by using the GMDH-type neural networks. In the paper, the GMDH-type neural networks are applied to the medical image recognition of the lungs.