Enhancing Invasive Ductal Carcinoma Detection in Breast Histopathology using Convolutional Neural Networks

Breast cancer is one of the major public health issues; due to this it is important to get diagnosed as soon as possible and correctly to enhance patient outcomes. Invasive ductal carcinoma is one of the common subtypes of breast cancer which is essential to be detected in the early stage for treatment. It is suggested a deep learning method employing Convolutional Neural Networks is applied to breast histopathology photos to address the issues with manual assessment and investigate the possibilities of automated IDC detection. The dataset compiled specimens of breast cancer having 162 whole-mount slides scanned at 40x magnification. The data retrieved 277,524 50 x 50 patches, 198,738 IDC-negative patches, and 78,786 IDC-positive patches from these slides. The patches are used to train the model by convolutional neural network and preprocessed. If IDC is found in the breast histopathology photos, then it is suggested to apply deep learning approaches. The performance of the automated method was demonstrated using a total accuracy of 94.78% including 93.12% precision, 94.23% recall, and 94.78% F1 score. This study is further used in breast cancer diagnosis using deep learning for automated Invasive ductal carcinoma. The help is provided to pathologists to identify invasive ductal carcinoma regions with a convolutional neural network. This further lowers the subjectivity in the results and raises precision in diagnosis. The results are helpful in automated IDC identification and improve treatment plans for patient care.