Kyocera leverages AI to spot skin cancer

Remote diagnosis seen helping with doctor shortage as Japan grays

Kyocera and the University of Tsukuba have developed a diagnostic system that taps the power of artificial intelligence to instantly detect possible skin malignancies from digital photos.

Deep learning was applied to review 4,000 images related to 14 types of conditions. For skin cancer, the system can assess the likelihood of malignancies from photos uploaded to servers by doctors. Tests using a general-purpose AI system from Kyocera Communication Systems achieved a diagnostic accuracy of roughly 90%.

The developers plan to enable the AI to also diagnose communicable diseases requiring a prompt response. Modifications may be made to make the system compatible with photos shot on tablets.

Kyocera and the university aim to provide a prototype to medical institutions in fiscal 2018 and begin sales the following year. A price has yet to be set, but they are looking to collect usage fees in addition to installation charges.

The ranks of specialist physicians in Japan are not growing fast enough to keep up with the expected rise in cancer and other patients as the population continues to age. AI will likely help fill the gap and enable even residents of areas with doctor shortages to receive early diagnoses and begin treatment right away.

Hitachi is gearing up to launch as early as autumn a system that aids physicians in diagnoses using CT and MRI images by calling attention to parts that may indicate illness.

A health ministry advisory panel encouraged the use of AI in medical image diagnostics and drug development in a June report. The ministry plans to add medical services employing AI to the government-set fee table as early as fiscal 2020 so that they are covered under the public health insurance system. It also plans to clarify rules regarding the use of AI in treatment under the medical practitioners’ law.

SOURCE: Nikkei