A new deep learning approach for automatic diagnosis of breast cancer

Breast cancer stands as a significant health concern, particularly for women worldwide. Regular breast examinations are crucial for the early detection and prompt treatment of this condition. Furthermore, computer-assisted diagnostic systems have made substantial strides in aiding pathologists during the diagnostic process. In our research, we introduce an innovative convolutional neural network for the accurate diagnosis of breast cancer from histopathological images. Unlike traditional convolutional neural network's that rely solely on raw image inputs, our model employs a dual-input architecture. One input utilizes the raw histopathological images, while the other leverages deep features extracted from related images. This dual-input approach enhances the network's ability to extract insights from additional information sources. We conducted all our experimental studies using the well-established BreakHis dataset. To assess the model's performance, we employed the accuracy metric and applied a rigorous 5-fold cross-validation technique. The results were remarkable, with accuracy scores of 99,94%, 98,94%, 99,05%, and 97,30% achieved for the 40×, 100×, 200×, and 400× sub-datasets, respectively. These results not only demonstrate the effectiveness of our proposed diagnostic system but also surpass the benchmarks reported in existing literature.