• Yash Vishwakarma
• Akhilesh A. Waoo

Breast cancer, medically referred to as cancer of the breast, is a highly fatal disease that results in numerous fatalities among women globally. Breast cancer screening is commonly performed using mammography and ultrasonography as the primary imaging modalities. Conversely, prevailing imaging modalities exhibit incapacity in discriminating between discrete subcategories of both benign and malignant neoplasms. Regarding this matter, histopathology-derived images may offer enhanced sensitivity in detecting both benign and malignant cancer types. The Vision Transformer has recently gained increased attention for medical imaging tasks due to its success in various computer vision tasks. The Swin Transformer model is founded on the concept of constraining the self-attention mechanism to shifted windows that do not overlap and has exhibited notable efficacy across various computer vision applications. This research aimed to assess the efficacy of the Swin Transformer model in categorizing eight subtypes of histopathology breast cancer images. The study utilized the BreakHis dataset, which comprises 7909 histopathological images captured at different zoom levels, namely 40X, 100X, 200X, and 400X. The Swin Transformer model was trained on the aforementioned dataset without utilizing any pre-trained weights, as is common in many Transformer studies. Despite this, the model achieved an average test accuracy of 94.05%, surpassing the performance of all the previous state-of-the-art works. Consequently, the Swin Transformer model tailored to the specific needs of the task at hand demonstrated proficiency in accurately classifying breast cancer subtypes through the analysis of histopathological images. This capability has the potential to alleviate the burden on pathologists.

swin transformer, histopathology images, multiclass classification, breast cancer classification, medical imaging

"SWIN TRANSFORMER FOR EXPLAINABLE BREAST CANCER CLASSIFICATION USING HISTOPATHOLOGY IMAGES", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.10, Issue 6, page no.a534-a541, June-2023, Available :http://www.jetir.org/papers/JETIR2306075.pdf

"SWIN TRANSFORMER FOR EXPLAINABLE BREAST CANCER CLASSIFICATION USING HISTOPATHOLOGY IMAGES", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.10, Issue 6, page no. ppa534-a541, June-2023, Available at : http://www.jetir.org/papers/JETIR2306075.pdf