An Efficient Neural-Fuzzy Weighted Multi Factor Decision Making Framework for Mammogram Mass Classification

Breast cancer is one of the fatal diseases of women. The survival rate can be improved with early diagnosis of the severity of the malignancy. In this paper we investigate multi criteria decision making system for classifying the mass and to find the severity of the tumour using fuzzy logic and neural network. Radiologists interpret the mass in mammograms based on three factors such as shape, margin and density. So it should be seen as a multidimensional problem involving multiple criteria when deciding the class of the mass and to scale the degree of malignancy. The inputs of the network are fuzzy discretizedto improve the representation of the features that describes the mass lesion characteristics as the radiologists do. Neural network is used to determine the fuzzy score of each criterion for the alternate classes benign and malign. Each criterion has been given weights and then weighted multi factor decision making framework is used to determine the class of the mass. The experiments using the proposed work have been implemented on Mammographic Image Analysis Society (MIAS) Database. The experiments were implemented in MATLAB.