Thermography-Based Early-Stage Breast Cancer Detection Using SVM

With the advances in health technology, disease diagnosis and detection are becoming more accurate and efficient. The use of Machine Learning (ML) algorithms has evolved health technology to not only provide detection tools, but also predictive analytics, clinical decision support, and treatment planning. Breast cancer being one of the most common forms of cancer among women, has resulted in the development of different diagnostic tools, such as ultrasound, mammogram, and biopsy. Thermography, which is a non-invasive technique that measures heat patterns on the skin surface can be used for early-stage breast cancer detection. With ML, thermography images can be used to identify any abnormality, which could indicate the presence of cancerous cells. In this study, Support Vector Machine (SVM) is used to detect abnormality in breast thermography. Pixel-based features are extracted from the images and then fed to SVM for high accuracy abnormality detection. The results are compared with other diagnostic tools, including, ultrasound and mammography. Experimental results show that the proposed solution achieves an accuracy of 68.90% and sensitivity of 95.45% by using features optimized with Principal Component Analysis (PCA) algorithms.