Invasive ductal carcinoma (IDC) detection in breast histopathology images using enhanced transfer learning of convolutional neural networks

Invasive ductal carcinoma (IDC) detection in breast histopathology images using enhanced transfer learning of convolutional neural networks

연구 분야: Artificial Intelligence

학회: International Journal of Information Technology

This study introduces a deep learning methodology for the automated identification of tissue regions indicative of Invasive Ductal Carcinoma (IDC) within Whole Slide Images (WSI) associated with breast cancer. Deep learning demonstrates notable efficacy in such applications, particularly when an ample number of samples are available for training purposes. The proposed framework extends across various convolutional neural networks (CNNs). The method underwent evaluation using a WSI dataset encompassing specimens from 162 patients diagnosed with IDC. The experimental outcomes indicate commendable accuracies of 89%, 88%, 84%, and 82%, for CNN, EfficientNet, EfficientNetV2, and VGG16, respectively.