Innovative infrared imaging approach for breast cancer screening: Integrating rotational thermography and machine learning analysis

Abstract

This paper presents a novel approach to breast cancer screening using infrared (IR) imaging. This work encompasses four phases: Refining data collection, advancing analysis methods, and enhancing feature extraction with machine learning. The developed system employed a temperature-controlled chamber with rotational thermography techniques to maintain consistent temperatures and capture high-quality IR images and all possible subject views. The paper describes four key experiments to detect breast cancer using IR imaging. The experiments involved the use of dynamic temperature-based data collection and a semi-circular arc movement to ensure precise imaging, keeping the object in focus. Initial experiments involved the use of dynamic temperature-based data collection and a semi-circular arc movement to ensure precise imaging focus. The final experiment incorporated a semi-circular arc movement. For each subject, 32 thermal IR images were acquired, targeting one breast at a time while isolating the other with an IR-proof barrier. The collected datasets were used for breast abnormality detection. The analyzed results revealed that support vector machine and neural network algorithms achieved an accuracy rate of 93.18%. The system’s installation at a hospital in India allowed for real-world application and validation. The final study, which introduced a new IR imaging protocol, demonstrated improved results compared to earlier pilot studies. This method enhances the accuracy of distinguishing malignant and benign tumors, supporting early breast cancer detection and treatment. The proposed methodology addresses data collection and analysis challenges, leading to improved screening efficiency and better patient outcomes.