Novel Supervised Learning Scheme for Optimizing the Classification Performance of Breast Cancer MRI

Abstract

Usage of machine learning has been always proven potential in identifying the best solution from the set of complex variables with the highly inter-twined relationship of problems. Similarly, supervised learning approach is one essential operation under machine learning that has always contributed in the area of healthcare and diagnostics. However, there are still some problems associated with the detection and classification of complex disease condition that is yet to be solved. The proposed system introduces a novel supervised learning approach along with a novel feature extraction scheme which is more progressive and less iterative. The proposed system considers a case study to perform classification of breast cancer using Magnetic Resonance Imaging (MRI) where it is subjected to normalization first followed by a novel segmentation process that compliments the classification operation too. The study outcome shows that the proposed system offers better classification performance in contrast to existing supervised approaches.