Validation of automated assessment (AI) of cervical images captured by Smart Scope® (SS) against colposcopy

Cancer of the cervix is a leading cause of death in Indian women and contributes to nearly one fourth of the world burden ofcervical cancer incidence and mortality. Access to screening and health services is also a major barrier for success in cervicalcancer screening programs. Pap smear test, a multi-visit screening test, is known to have low sensitivity. As per therecommendations of R. Sankarnarayanan for cervical screening in low resource setting, Visual Inspection with Acetic Acid (VIA)followed by Visual Inspection with Lugol’s Iodine (VILI) is a sufficiently sensitive and specific technique when performed bytrained and semi-skilled paramedical professionals.

Periwinkle Technologies has developed the digital viewing device “Smart Scope®â€. Smart Scope’ is a portable, electricityindependent device which offers ‘single visit test result’ at an affordable cost in a primary healthcare OPD setup. This device further enhances the efficacy of VIA-VILI test by allowing close examination (3 to 4cm) of the cervix, maneuverability, digital logkeeping and remote consultation. SO far we have used Smart copy assessment as a screening tool for cervix for more than 1500 women (data yet to be published). Now, with the desire to improve the efficiency of clinical care, we have introduced Artificial Intelligence (AI) based MachineLearning enabled auto-assessment feature in the software platform of Smart Scope®. The auto-assessment feature will classifyecto-cervix into 4 categories viz., normal (green) / abnormal benign changes (amber) / High Risk Amber (HRA)/ pre-cancerous & cancerous changes(red). This will negate the subjectivity from interpretation of lesions and remove the dependency on trained manpower at primaryhealth centers.

Many studies have been undertaken to evaluate AI performance in oncology, its application in clinical use and cancer care. Wehave developed (trained and validated in lab) one such Machine Learning (ML) Model. . This is an observational, non-randomized single arm trial to be conducted at a single centre in India for validatingthe artificial intelligence of cervical image captured by Smart Scope against Colposcopy. The study will enroll 150 women and is expected to continue for 12 months. The efficacy of this classification willbe verified with colposcopy for all and histopathology as a gold standard, where ever applicable.