India accounts for one-third of the global incidence and mortality of oral cancer. The national oral cancer screening program uses Conventional Oral Examination by Primary Healthcare (PHC) workers. This subjective assessment leads to unnecessary referrals to higher centers for biopsy and false negatives due to inappropriate biopsy site selection. Angiogenesis is one of the steps in early carcinogenesis, as solid tumors, such as oral cancers, cannot grow beyond 2-3 mm in diameter without inducing their own blood supply. The increased vascular supply and metabolic rate in malignant cells lead to a rise in temperature, which can be detected by sensitive digital infrared (IR) cameras. An Artificial Intelligence-enabled automatic analysis of intraoral IR images of oral lesions can be used for screening, early detection of Stage I and II oral cancers, and biopsy site selection as an objective Point-of-Care (POC) adjunct. A standardized protocol for passive and active IR imaging of intraoral lesions (index test) with a smartphone-based IR camera will be prepared to train a Medical Scientist and Technician. IR images of already diagnosed normal, Inflammatory, potentially malignant disorders, and malignant oral lesions (N = 100 each) will be used in Phase I to train an AI model to classify images as malignant or non-malignant. A second set of IR images of these oral lesions will be used in Phase II (N = 100 each) for evaluating the performance of the AI model. The reference test for malignancy will be histopathology (Gold standard). The intra/inter observer reliability will be assessed using the kappa statistics A clinical trial and validation of the proof of concept are proposed for IR imaging of intraoral lesions as a noninvasive POC adjunct for early detection and screening of oral cancer by PHC workers, and for the selection of biopsy sites by surgeons. © 2026 Wiley Periodicals LLC.
Gupta et al. (Wed,) studied this question.