Artificial Intelligence (AI)-Deep Neural Network (DNN) based Classification model to find BI-RADS score from Mammogram Images
Snehal C. Sapkale, Swapna Yenishetti, Lakshmi Panat · 2024
Breast cancer is the leading cancer among women globally. It is reported that with every four minutes, an Indian woman is diagnosed with breast cancer. To describe cancer, concept of staging is used. The cancer's stage tells us how far it has grown into nearby tissues. In breast cancer, there are total five stages from 0 (zero) to 4. If breast cancer is detected early between the stage 0-2, then its five-year survival rate exceeds 90% [1]. Survival becomes more difficult in higher stages of cancer. The low survival rates of breast cancer in India is due to delay in detection. Breast imaging tests - such as mammograms, magnetic resonance imaging (MRI) and ultrasounds, help doctors assess breast tissue for the early detection of breast cancer. The results of these tests are scored as BI-RADS (Breast Imaging Reporting and Data System) categories. These categories ranges from category 0 to category 6 (high likelihood of cancer). Radiologists determine the BI-RADS category by applying specific criteria and assessing the imaging tests. There is a need of an efficient and reliable tool that would help in automatic reading of mammogram characteristics and also assist radiologists in mammographic image interpretation. This paper describes a deep neural network (DNN)-based classification model that would efficiently detect mammogram characteristics and categorise it into a relevant BI-RADS category with an accuracy of 71.78%.