Multi-Input Deep Learning Approach for Breast Cancer Screening Using Thermal Infrared Imaging and Clinical Data

Breast cancer is one of the most prevalent causes death among women across globe. Early detection best strategy for reducing mortality rate. Currently, mammography standard screening modality, which has its shortcomings. To complement this thermal infrared-based Computer-Aided Diagnosis (CADx) tools have been presented as economical, less hazardous, and a suitable solution various age groups. Although viable solution, CADx systems are built primarily from frontal breast thermograms, likely to miss lesions that may develop on sides. Additionally, these often disregard critical clinical data, such risk factors. This paper presents novel system utilizes deep learning techniques detection. The incorporates multiple thermogram views corresponding patient data improve accuracy diagnosis. We describe methodology system, including extraction regions interest images use transfer train three different models. evaluate performance models compare them similar works literature. results demonstrate using multi-inputs outperforms single-input achieves an overall 90.48%, sensitivity 93.33%, AUROC curve 0.94. approach could offer more cost-effective hazardous option detection, particularly wide range