Oral Cancer Screening using AI

The research activity involved the development of an AI model embedded within a mobile app that can detect subjects with a high risk of transformation into malignancy by processing white light images of the oral cavity captured from smartphone cameras, thereby developing a point of care, AI-assisted, phone-based diagnostic device, ensuring availability, accessibility and affordability. To achieve this, both CNN and transformer-based models were experimented with such as Inception, Vision Transformer and a few other lightweight models such as EfficientNet, MobileViT, and FastViT. Image quality assessment tools such as NIQE, CLIP-IQA, segmentation tools like bounding boxes, and SAM were also experimented.