Combination of Gabor Filter and Convolutional Neural Network for Suspicious Mass Classification

Abstract

In this paper, we present an innovative method for classifying suspicious regions in mammography images into massive or normal ones. We consider Gabor filter as a processing step to make the feature of masses in mammography images clearly. Then we use convolutional neural network to train and test the detected masses from real data sources in Mammographic Image Analysis Society (MIAS) and Digital Database for Screening Mammography (DDSM). The result is quite sanguine when compared with other approaches. We achieve the sensitivity of 96% in 2700 regions detected from MIAS and 1000 regions extracted from DDSM.