Automated Detection of Breast Cancer in Histopathology Images Using Convolutional Neural Networks

In this work, we classified IDC-positive and IDCnegative patches from breast histopathology pictures using an automated detection method built on Convolutional Neural Networks (CNNs). A CNN model intended to detect malignant areas in breast tissue samples was trained using a dataset including 277,525 patches (198,738 IDC-negative and 78,786 IDC-positive). With a macro-average F1-score of 0.90 for both classes the model attained an overall accuracy of 95 %. But a somewhat lower F1-score of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{0. 8 3}$</tex> for pictures positive of IDCpositive suggests the difficulty in spotting cancerous areas brought on by class imbalance. Future research will concentrate on class imbalance resolving and innovative data augmentation methods enhancing IDC identification. These findings show how CNNs could improve automated techniques of breast cancer detection.