SwinCNN: An Integrated Swin Transformer and CNN for Improved Breast Cancer Grade Classification

Breast cancer grade classification plays a critical role in determining appropriate treatment strategies. This study introduces SwinCNN, a novel integration of Swin Transformer and Convolutional Neural Networks (CNN) aimed at enhancing breast cancer grade classification accuracy. The process begins with preprocessing of input images, including resizing to 256x256 pixels and histogram equalization for normalization. Features are extracted from the resized and normalized images using the ResNet-18 CNN, followed by additional feature extraction through a Swin Transformer Block to capture both global and local patterns. These extracted features are then combined and classified using a Support Vector Machine (SVM) with a linear kernel. The approach demonstrates improved classification performance by leveraging the complementary strengths of both CNN and Swin Transformer, offering significant advancements in breast cancer grading. Results indicate a high classification accuracy, underlining the effectiveness of the integrated model in medical imaging applications. This integrated framework promises to refine diagnostic processes and support precise treatment planning in clinical settings.

Keywords: Breast Cancer Dataset, Features Extraction, Image Processing Techniques, Swin Transformer, Machine Learning Algorithm, Deep Learning algorithm, and Accuracy.