Clustering and Least Square Based Neural Technique for Learning and Identification of Suspicious Areas within Digital Mammograms

This paper presents a technique which explores the fusion of clustering and a least square method for the classification of suspicious areas within digital mammograms into benign and malignant classes. It incorporates a clustering algorithm such as k-means in conjunction with a gram-schmidt based least square method. The main focus of the research presented in this paper is to (1) improve the classification of features from suspicious areas within digital mammograms and (2) examine the effects that the determined clusters and least square methods have on classification accuracy and efficiency. The proposed technique has been tested on a benchmark database and the results from preliminary experiments are discussed.