Point of Care Detection and Diagnosis of Oral Cancer using a Low Cost Imaging Module enabled by AI

PROJECT SUMMARY/ABSTRACT The overall vision of the proposed project is to to develop and deploy an affordable automated point-of-care (POC) telecytology platform for oral cancer screening that will reliably establish a diagnosis of oral cancer in the community setting and establish an immediate referral care pathway. Oral cancer is a significant public health problem in India; 77,000 new cases and 52,000 deaths are reported annually, which is approximately one-fourth of global incidences. Approximately 70% of cases present at an advanced stage, when the probability of cure is very low, and a five-year survival rate is around 20%. It has been estimated that early diagnosis, with timely and proper treatment, could improve the survival rate up to 90%. The current ‘gold standard’ of oral cancer screening is visual inspection of the mouth by trained individuals, followed by biopsy of suspicious lesions. However, in India there is a delay of nine months from the onset of symptoms to diagnosis. Of this, seven months are attributed to the delays within the medical pathway The majority of the population lives in a rural environment, where access to pathology services and expertise is very limited. Without definitive proof of cancer, patients are not eligible for state-run insurance programs for treatment. Our proposed approach comprises a portable system for scanning brush biopsy cytology slides with cloud connectivity for transmission of images to pathologists and/or automated diagnosis via a validated algorithm for identification of atypical cells. After standard visual triaging of patients during routine screening, those identified with higher risk lesions will immediately be directed to undergo brush biopsies on the same day. Samples will be placed on a glass slide, stained with routine toluidine blue (average time is <4 minutes), and imaged using the portable slide scanner. Initially these images will be relayed via cloud to a remote pathologist who will immediately report them, while subsequent versions of the prototype will have in-built artificial intelligence (AI) algorithms for automated reporting in the field. We believe that this innovative and affordable workflow would successfully expedite diagnosis and provide significantly earlier treatment for oral cancer patients.