Breast Cancer Detection using DCNN and MSVM

Abstract

Breast cancer screening is a critical area of medical diagnostics, where the accuracy and performance of radiologists play a pivotal role in early detection and diagnosis. In this Project, we present a novel approach aimed at enhancing radiologists performance in breast cancer screening through the optimization of parameters for a Multi-Class Support Vector Machine (MSVM). We compare the results of our proposed method against an existing approach based on Deep Neural Networks (DNN) in terms of accuracy, specificity, and the types of cancer detected, including both benign and malignant cases. The existing method employs DNN as the primary algorithm, achieving an accuracy rate of 92.8%. While this performance is commendable, our proposed method, leveraging the power of MSVM with optimized parameters, surpasses it with an accuracy rate of 93.5%. The proposed DCNN architecture is designed to automatically learn discriminative features from mammography images. The model consists of multiple convolutional layers followed by pooling layers to extract hierarchical features. Additionally, batch normalization and dropout layers are incorporated to improve the generalization and robustness of the model.