Doctors design low-cost AI cervical cancer detection model to screen samples in 30 secs

Kolkata: A group of doctors and researchers from 4 institutes in Kolkata conducted a study on an AI-enabled, low-cost cervical cancer screening model designed for rural and resource-constrained settings and found high diagnostic accuracy. This AI model, using Android device-captured images of smear samples, demonstrated that it took 30 seconds at the most to screen a sample. This study was published recently in the Asian Pacific Journal of Cancer Prevention.Experts said the conventional routine screening procedure for cervical cancer at present is a Pap smear, also called the Pap test. This test can detect abnormal or precancerous lesions by examining cells taken from the cervix under a microscope. While this screening could take less than 30 seconds, analysing the samples in the lab took longer. The average cost of a Pap smear is around Rs 1,000 in Kolkata. The screening cost could come down to as low as Rs 200 using this AI software.
For the study, an AI programming software was developed, with an integrated system that automatically detected and classified cells from microscopic Pap smear slide images taken on Android phones or tabs to diagnose cervical cell morphology in a time-efficient and cost-effective manner.A total of 292 smear samples from hospital-based labs were considered, from Calcutta National Medical College (CNMC) and NRS Medical College. The Pap smear images were analysed through the newly developed AI software, using the required standardisation and validation method.The study found that the custom AI model could successfully classify 98.1% and 80.5% of normal and abnormal cells in hospital in-house samples, respectively. In addition, the team also found a significant correlation between biopsy, which is the gold standard confirmatory test, and AI reports."This AI-based cervical screening is very promising, especially in rural settings that see higher cervical cancer prevalence but lack of diagnostic facilities," said pathology associate professor Arindam Karmakar, who was then posted at CNMC when the study was conducted.Priyabrata Das, scientific director, CliniMed Lifesciences, who was in the research team, said that screening using this AI tool took less than 30 seconds and the accuracy rate was high.The study, titled ‘Evaluation of the Diagnostic Accuracy of Cervical Cell Morphologies from Android Device-Captured Cytopathological Microscopic Images through Artificial Intelligence in Mainly Rural or Resource-Constraint Areas of India', concluded that while manual screening of cervical cytology smears is time-tested, AI is set to revolutionise the process by improving outreach, availability, accuracy, and economy. "If implemented responsibly, ethically, and with continued validation, such AI-driven tools can significantly accelerate India's progress towards reducing cervical cancer mortality," said Snehendu Konar, clinical operations lead, Auriga Research. In addition to Karmakar and Das, the CliniMed LifeSciences team, with Arindam Ray, Soumyabroto Banerjee, Swati Dasgupta, Sayan Dey, Umar Faruk, worked together with Moumita Maiti from NRS and Prosenjit Saha from CNCI Kolkata.