Untitled

Abstract

Breast cancer has been a major challenge in the arena of health sector as reports from literature quote a 20% of women population being affected by this fatal disease on a global scale and nearly 32% of it being shared by Indian population. However, if detected and treated at an early stage, there could be a massive reduction in the fatality rates. In recent time, medical health sector has achieved great milestones in detection and treatment of various disorders, thanks to the rapid pace of research and advancements in various fields like imaging modalities, segmentation and 3D processing techniques and availability of a powerful set of classifiers and predictors given a data set. Machine learning models have recently evolved as a powerful tool providing solutions to a number of such research challenges that involve intelligence, learning decision making with pinpoint accuracy. This work proposes a deep convolution neural network model -VGG-19 to attain high accuracy and precision in the cancer detection process. The computational complexity is effectively reduced by the proposed model over conventional methods the proposed deep learning model is able to perform both preprocessing and feature selection. Along with a powerful classifier inbuilt, an SVM is integrated into the proposed model to enhance the detection rate. The work has been carried out using the Wisconsin Breast Cancer Dataset (WBCD)which has an expansive data set of mammographic images. Performance metrics have been validated and justified by comparing against recent methods in terms of metrics like detection accuracy, precision, recall.