Grade Classifcation of Breast Cancer using Deep-Learning

Breast cancer(Bcancer) in India has a high mortality rate, with one woman dying for every two newly diagnosed. The primary reasons for this are delayed diagnosis, advanced stages of the disease upon presentation, and inadequate facilities for early detection and treatment. Bcancer does not often show early symptoms, making it difficult to prevent. Early detection can significantly improve the prognosis and survival rate by allowing for timely clinical treatment. Automating some elements of Bcancer diagnosis can increase the number of women who are screened and detected early, ultimately reducing the number of late-detected cases. By combining traditional and computer-based diagnosis methods, pathologists' workload can be reduced, and performance can be improved. Identifying and classifying subtypes of Bcancer accurately is a crucial clinical task, and automation can save time and reduce errors.further,Two models have been created for classifying histopathological images into grade I (G1), grade II(G2), and grade III(G3) BCancer. One model is designed for binary classification, while the other model is designed for multi-class classification.Despite limited computational power and a small dataset, the results are satisfactory. The models can be improved for grade-wise classification of Bcancer. Deep networks have been designed with computational tricks such as ReLU, dropout, and batch normalization to improve performance. There are several algorithms to detect breast cancer, but there is currently no grade classification for breast cancer. In this paper, our focus is on classifying the different grades of breast cancer, which will help patients understand the severity of their condition. We classify G1 as the least dangerous stage, G2 as more dangerous, and G3 as the most dangerous stage.