Enhancing Breast Cancer Detection Using A Transformer-Based Model

Breast cancer is a very common and one of the leading causes of mortality and illness worldwide. Early detection of breast cancer is critical, increasing the chances of long-term survival and successful treatment. Mammography is a primary method for early detection and a widely used imaging technique employing low-dose X-rays. Interpreting mammograms remains challenging due to the complex nature of malignancy anomalies. This advanced imaging technique has revolutionized the screening of breast cancer by facilitating the better visualization of breast tissues. The main challenge that radiologists face in reading mammographic patterns is too many false positives and negatives, which leads to unnecessary biopsies, anxiety among patients, and missing the correct diagnosis. In current times, artificial intelligence (AI) has shown significant success in the medical imaging field. In particular, deep learning (DL) models have outperformed traditional image processing techniques. Allowing accurate and automated predictions, these models are trained on massive datasets to understand the different characteristics of tumors. In this work, a different DL-based model is deployed and developed to detect breast cancer using mammogram images. Also, several DL approaches are discussed, mentioning their benefits and limitations concerning mammography. This work thoroughly analyzes the literature and uses techniques and different evaluation methods to analyze the model's performance across different datasets.