Breast Mammograms Diagnosis Using Deep Learning: State of Art Tutorial Review

Usually, screening (mostly mammography) is used by radiologists to manually detect breast cancer. The likelihood of identifying suspected cases as false positives or false negatives is significant, contingent on the experience of the radiologist and the kind of imaging screening device/method utilized. The confirmation of the type of tumour seen by the radiologist is sent for histological investigation (microscopic analysis) through a biopsy, where the tumor's grade and stage, which are used in the latter stages of treatment, are ascertained through biopsy. However, a secondary issue with the cancer detection process is that only 15 to 30% of instances that are referred for biopsy result in malignant findings. Since deep learning demonstrated remarkable performance in visual recognition challenges, it has been widely applied to a variety of tasks. Similar examples include deep learning applications in healthcare, which are gaining a lot of interest from the research community. Deep learning is used to identify, categories tumours, and breast cancer is a significant global health concern. The medical sciences could now make more accurate diagnoses and detections due to recent advancements in machine learning techniques. Hence due to systems potential accuracy, it could offer optimistic outcomes when used to read malignant images. In imaging domains, deep learning-based methods have achieved remarkable success in constituent segmentation (UNet), localization (DenseNet), and classification (VGG-19). This study examines, how deep learning methods are assisting in the highly accurate diagnosis of benign or malignant tumours based on screened images. In contrast to a mammogram, which is covered in detail, this paper briefly discusses imaging methods for cancer detection. Early detection and cost effectiveness are two main benefits of applying machine learning and deep learning techniques to mammograms.