Breast Cancer Classification Using Fine-Tuned SWIN Transformer Model on Mammographic Images

Breast cancer is the most prevalent type of disease among women. It has become one of the foremost causes of death among women globally. Early detection plays a significant role in administering personalized treatment and improving patient outcomes. Mammography procedures are often used to detect early-stage cancer cells. This traditional method of mammography while valuable has limitations in its potential for false positives and negatives, patient discomfort, and radiation exposure. Therefore, there is a probe for more accurate techniques required in detecting breast cancer, leading to exploring the potential of machine learning in the classification of diagnostic images due to its efficiency and accuracy. This study conducted a comparative analysis of pre-trained CNNs (ResNet50 and VGG16) and vision transformers (ViT-base and SWIN transformer) with the inclusion of ViT-base trained from scratch model architectures to effectively classify mammographic breast cancer images into benign and malignant cases. The SWIN transformer exhibits superior performance with 99.9% accuracy and a precision of 99.8%. These findings demonstrate the efficiency of deep learning to accurately classify mammographic breast cancer images for the diagnosis of breast cancer, leading to improvements in patient outcomes.

Improving Breast Cancer Diagnosis with SWIN Transformer Models

What is it about?

This study focuses on improving breast cancer diagnosis using advanced machine learning techniques. Breast cancer is one of the most common cancers among women, and early detection is crucial for effective treatment. Traditional mammography methods, while useful, can result in false positives or negatives, leading to delayed or unnecessary treatments. The research compares two popular deep learning approaches: Convolutional Neural Networks (CNNs) and vision transformers. Specifically, it evaluates models like ResNet50, VGG16 (CNNs), and SWIN transformer and ViT-base (transformers) on mammographic images to classify tumors as benign or malignant. The study uses a large dataset of over 24,000 images and fine-tunes these models to optimize their performance. Among all tested models, the SWIN transformer outperformed others, achieving near-perfect accuracy (99.9%) and precision (99.8%). The study highlights how vision transformers, with their advanced feature extraction capabilities, are better suited for complex medical imaging tasks. It also emphasizes the importance of using pre-trained models for consistent and accurate results.

Why is it important?

Breast cancer remains a leading cause of death worldwide. Early and accurate detection can significantly improve survival rates and reduce the burden on healthcare systems. Traditional diagnostic methods like mammography often rely on human interpretation, which can be time-consuming and prone to errors. Machine learning offers a way to automate and enhance this process, reducing false diagnoses and improving efficiency. This research is important because it demonstrates the potential of vision transformers in medical imaging. The SWIN transformer model, in particular, shows exceptional ability to detect subtle patterns in mammographic images that may be missed by other methods. This advancement could lead to earlier and more accurate breast cancer diagnoses, reducing the need for invasive procedures like biopsies. Additionally, this work provides a blueprint for integrating machine learning into routine clinical workflows, potentially improving outcomes for patients and making healthcare more cost-effective. The findings can inspire further research into applying vision transformers for other types of medical imaging.

Key Takeaways:

1. SWIN transformer achieved 99.9% accuracy in classifying breast tumors.

2. Vision transformers outperform traditional CNNs in breast cancer detection.

3. Advanced models reduce false positives and unnecessary biopsies.

4. Pre-trained models provide reliable and scalable diagnostic tools.

5. This approach could improve early detection and save lives globally.