Evaluation of an AI-powered Portable Thermal Imaging Solution as a Pre-screening Tool for Breast Cancer

Abstract

Background and objectives: Breast cancer is one of the greatest global health concerns for women, with rising incidence rates and mortality projections, while affordability and access to mammography screening and diagnosis, especially in lowand middle-income countries, remain a challenge. This retrospective clinical validation study evaluated a breast cancer prescreening solution (BCPS) based on a commercially available smartphone with a thermal imaging sensor powered by artificial intelligence. The purpose was to measure the performance of the BCPS tool compared to mammography, the gold standard for first-pass examination in breast cancer screening. Methods:The evaluation was conducted in the Erebouni Medical Center Breast Unit in Armenia over a period of six months. We tested a cohort of 478 women of whom 45 were finally diagnosed with breast cancer after biopsy. Participants were first screened with the BCPS before undergoing the standard breast screening pathway. After studying the mammography results, if malignancy was discovered, a biopsy was performed and taken as the ground truth when comparing with BCPS artificial intelligence results. Results:When combined with patient-reported or clinical symptoms, the BCPS tool achieved a sensitivity of 89% and a specificity of 83% compared to mammography. When clinical or patient-reported symptoms were not taken into account, sensitivity was considerably lower (60%), while specificity was higher (88.2%). Conclusions:The BCPS tool, in combination with basic clinical exams and patient-reported symptoms, may serve as a robust triaging tool for breast cancer detection where mammography is not available or affordable, identifying the majority of women who need further diagnostic assessment.