Screening mammogram images for abnormalities using radial basis Function Neural Network

Intelligent Computer Aided Diagnosis (CAD) Systems can be used for detecting Microcalcification (MC) clusters in digital mammograms at the early stage. CAD systems help radiologists in identifying tumor patterns in an efficient and faster manner than other detection methods. In this paper, we propose a new approach for detecting tumors in mammograms using Radial Basis Function Networks (RBFNN). Prior to the detection of MC clusters features from the image are extracted and analyzed. Gabor features are extracted from the image Region of Interest (ROI) to distinguish a tumor cluster and a normal breast tissue. Once the features are extracted, they are given as input to the supervised RBFNN. The output neuron determines whether the given input ROI is cancer tissue or not. We have verified the algorithm with 322 mammograms in the Mammographic Image Analysis Society Database (MIAS). The results shows that the proposed algorithm has a sensitivity of 85.2%.