A Novel Neuro Fuzzy Deep Random Vector Functional Link Network for Early Cancer Diagnosis Using Thermography

Breast cancer ranks among the most prevalent cancers impacting women globally. Regular screening plays a crucial role in early detection, helping to lower risks and improve survival outcomes. While mammography is a widely adopted screening tool, its limitations such as exposure to radiation, reduced sensitivity in dense breast tissues, and high costs necessitate exploring alternatives. Thermography is a non-invasive and radiation free medical imaging technique that measures heat patterns on the surface of the skin, to detect abnormalities that indicate the presence of breast cancer. Despite the advantages, its effectiveness is often limited by subjective interpretations and low sensitivity in detecting small or deeply located tumors. This paper addresses these challenges by proposing a neuro fuzzy deep random vector functional link network for early detection of cancer from thermal images. The architecture is designed for fast training and adaptability, with the fuzzy layer enhancing both interpretability and robustness. Together, they create a hybrid model that effectively balances accuracy, computational efficiency, and the ability to manage imprecise data. With performance metrics of 96.44% accuracy, 95% sensitivity, and 98% specificity, the model demonstrates significant potential as a superior alternative for breast cancer screening. By integrating interpretability with advanced learning capabilities, the model contributes significantly to the growing field of AI-driven diagnostic tools, offering a promising solution to a pressing global health challenge.