Analysis method of mammogram based on neural network with fuzzy reasoning

An analysis method for mammograms based on a neural network with fuzzy reasoning is presented. In order to analyse mammograms the image preprocessing is very important. Here, the authors study stripping the tumors from the mammogram based on the principle of maximum gray threshold and the parameters of the tumor are extracted from the tumor. Meanwhile, the membership function of feature parameters and fuzzy rules are deducted according to expert´s knowledge. The rules are used as input parameters of the hybrid neural network which is composed of a fuzzy reasoning cooperative network and a BP net. Finally, four classes of breast disease are classified from the mammograms by the hybrid neural network