MSTLA: Multi-Stage Transfer Learning Approach for Breast Carcinoma Diagnosis
Gunjan Chugh, Shailender Kumar, Nanhay Singh · 2023
In the recent era, Breast Carcinoma has been observed as a fatal disease in women. In India and other developing countries, survival rates are very low because detection occurs very late. Computer-aided diagnosis(CAD) has emerged as a tool for helping doctors and experts with early detection and diagnosis. Medical Image Processing relies upon processing medical images to segment and categorize several diseases. Machine Learning(ML) and Deep Learning(DL) have shown tremendous success in recent years in various real-life applications, including speech recognition, face recognition, autonomous vehicles, etc. Transfer Learning is a technique where features learned from one domain are transferred to another. This approach is generally followed when the data required for training a model is unavailable in abundance. In this work, we have designed a model for the early diagnosis and categorization of breast malignancy using the Multi-Stage Transfer Learning Approach(MSTLA). DenseNet169 and ResNetl52 are utilized for three stage transfer learning strategy. The results show that both DenseNet169 and ResNetl52 performed remarkably with an accuracy of 100% and 99% in the third stage of transfer learning.