A parallel computer system for the detection and classification of breast masses

Abstract

Breast masses are regarded of paramount importance attributed clear malignancy targets whose detection proves acute for breast cancer diagnosis. Despite the boost in technology that has enhanced diagnostic and clinical developments in the medical field; the accuracy in mammographic tumor evaluation still remains a comprising issue. This is with an even greater regard towards mass detection. This paper is aimed towards creating a dynamic mammographic image enhancement system in parallel to a tumor detection and classification system. A dynamic list of histogram based algorithms constitutes the enhancement system. The detection and classification system comprise of a Seed Region Growing (SRG) segmentation algorithm and a Multi Layer Perceptron (MLP) neural classifier using the Backpropagation algorithm. Results have rendered the proposed techniques promising with accurate levels of benign and malignant tumor discrimination and enhanced breast image quality. The latter system achieved 88% sensitivity, 72% specificity, an Az value of 0.84 and an overall classification accuracy of 80%.