Microsoft is working with SRL Diagnositics, an Indian pathology and radiation diagnostic service company, to reduce the time it takes to diagnose cervical cancer with AI. In India, 67,000 women die each year from cervical cancer. It accounts for more than 25% of cervical cancer deaths worldwide.
Both sides included an AI trained using thousands of images of already diagnosed cervical cells to detect cancer as well as detect cervical development with precancerous lesions. This way, rather than the doctor looking for a lesion by comparing it to a single picture of a sample, the AI can check whether the image that has been determined in advance is correct or incorrect.
The AI developed this time is currently in beta and is not yet used for diagnosis. However, it is said that it has a performance that is quite close to practicality compared to AI diagnostic tools that are being tested in other fields. As stated earlier, in India, countless women die every year from cervical cancer. But at the same time, the lack of willingness to diagnose it is also a problem. According to reports, SRL Diagnostics diagnoses more than 100,000 samples every year, but 98% of them are considered normal. In order to detect the remaining 2%, the doctor has to look at all of the mass images.
If AI can allocate the first screening task, it is expected that the number of images that doctors need to diagnose will be drastically reduced, and the waiting time until the diagnosis result will decrease. Related information can be found here .