Breast Cancer Detection Using Deep Learning and Feature Decision Level Fusion

Among women, breast cancer has a high incidence and high fatality rate. Due to a lack of early detection facilities and barriers to accessing technological improvements in battling this illness, mortality rates are disproportionately greater in underdeveloped countries. Biopsies done by trained pathologists are the only certain approach to diagnosing cancer. With the use of computer-aided diagnostic algorithms, pathologists may improve their efficiency, objectivity, and consistency in making diagnoses. A key goal of this research is to create an accurate automated system for diagnosing breast cancer that can function in the current clinical setting. In this work, we offer an algorithm for the identification of breast cancer that uses asymmetric analysis as the basic choice and decision-level fusion. Fusion of local nuclei features extracted using convolutional neural network (CNN) models pre-trained on the database constitutes the picture feature representation. The dataset is accessible for public use, and the results are evaluated by running 25 random trials with an 80%-20% split between train and test. Overall, the suggested framework was 86%. The proposed framework is shown to outperform numerous current methods and to provide results on par with the state-of-the-art techniques without requiring extensive computing resources. Breast cancer detection from histological pictures may be greatly aided by the use of this qualitative approach based on transfer learning.