BC-MBINet: A Novel Architecture for Accurate Classification of Breast Cancer with Microscopic Biopsy Images Using Deep Convolutional Neural Networks
Kuljeet Singh, Amrit Sudershan, Sachin Kumar, Sourabh Shastri, Vibhakar Mansotra · International Journal of Pattern Recognition and Artificial Intelligence · 2025
Breast cancer (BC) is the second most frequent malignancy, accounting for roughly 25% of all cases of cancer. BC is caused by genetic, epigenetic, and environmental factors, and their interaction too. The diagnosis of a BC is a critical step in the treatment process, and histopathological imaging is required to determine the type of illness. Identifying a disease is an important stage in the treatment process. However, this time-consuming task is exhausting, and people are prone to making mistakes that go unnoticed, making it difficult to determine the severity of the condition and this diagnosing step also relies on a pathologist’s expertise. In this paper, we have developed a novel BC with microscopic biopsy images network (BC-MBINet) model using deep convolutional neural networks. Feature extraction is handled by a sequence of convolutional layers, nonlinearity is handled using LeakyReLU activations, and learning is stabilized by batch normalization. A last Softmax layer is employed for binary classification into benign and malignant tumors, and dropout layers are included to decrease overfitting. The model achieves state-of-the-art accuracy and resilience in discriminating BC types by being trained on a publically available dataset of microscopic biopsy images. The proposed model is capable of classifying between the benign and malignant BC tumors with 99.04% accuracy. The model gives state-of-the-art results in its accuracy in classifying BC tumors into Benign or Malignant.