Breast Cancer Segmentation & Classification of Image using ResNet

Abstract

Cancer is one of the deadliest disease, which leads living things to death, and there are nearly 18 million cases, which includes 9.5 million men and 8.5 million women. Breast cancer is one of the most common cancer globally in women. Considering such devastating statistics of breast cancer, early detection is needed, in past several researcher have tried to detect in the early stage, however the main disadvantage of these models were its complexity since the detection comprises many phases such as segmentation and classification. Furthermore classification plays an eminent role in early detection as it detects the cancer type i.e. benign or malignant. In this paper, we have proposed a novel classifier M-ResNet (Modified ResNet) to classify the cancer types. At first, we develop a learning framework and selects the optimal kernel. Further, we employ the M-ResNet (Modified ResNet) and classify the cancer based on its Bi-Rads score. Furthermore, our model is evaluated by considering the performance metric such as Accuracy, sensitivity and specificity. Our model achieves the accuracy of 96.43.