Breast Cancer Detection on Histopathological Data

The leading cause of cancer-related fatalities among women worldwide is breast cancer, which is a severe public health concern. The likelihood of survival can be considerably increased by early identification. Although the accurate diagnosis of malignant tumors from histopathological pictures is mostly dependent on radiologists' long-term expertise, specialists often disagree with their decisions. Computer-aided diagnosis adds a second option for visual diagnosis, potentially increasing the validity of professionals' decisions. Automatic and precise classification of breast cancer histopathological images is critical in clinical applications for identifying malignant tumors from histopathological images. Deep learning has achieved excellent results in several disciplines such as image classification, object recognition, and so on. Deep learning for the analysis of medical pictures has recently attracted the attention of numerous researchers. In this field, convolutional neural networks (CNNs) have become a well-known class of models. To this end, the paper proposes using the well-known CNN architectures ResNet50, AlexNet, and GoogleNet which have been trained from scratch, in conjunction with various preprocessing techniques such as Contrast Limited Adaptive Histogram Equalization (CLAHE), Image sharpening, Edge enhancement, and so on, for the task at hand. The optimum model amongst the proposed i.e., Resnet50 + Image Sharpening achieves an accuracy of 95%.