AI Assisted Medical Fundus Detection | Accuracy has reached over 90%

Due to the lack of ophthalmology staff in Taiwan and the low referral rate of diabetic eye tests, ITRI has developed the "AI Recognition Medical Imaging System." Based on a large database of fundus images and a database of physician identification marks, AIU assisted physicians in screening High-risk groups and patients in need of care, to improve the prevention and treatment, the current accuracy has reached more than 90%

Chen Jianren, manager of ITC's Intelligent Clinic System, pointed out that due to the current shortage of ophthalmologists in Taiwan, there are fewer than 1,700 ophthalmologists in Taiwan. However, 35% of patients with diabetic mellitus experience retinopathy after the onset of 10 years, Will lead to further blindness.However, diabetic patients who are accustomed to seeing a doctor's home or a general medical department in a district hospital or district clinic have only confirmed that 17.4% of them have a referral to an eye clinic when they have any doubt of fundus diseases. Therefore, Equipment, hoping to help home medical doctors with artificial intelligence eye fundus examination.

The AI-Recognition Medical Imaging System uses TensorFlow, the Goole open-source machine learning system, to conduct thousands of in-depth studies with 50,000 images from open databases and has undergone another 50,000 images tests, the accuracy of which has now reached 90 %. The device has entered the clinical validation phase, hoping to start in 2018 the device into the home medical clinic.

In Taiwan, there are as many as 2.27 million people with diabetes, and the overall rate of myopia in the country is as high as 43%. There is a high risk of macular degeneration and cataracts. The unit specifically shows the relationship between ITRI and regional hospitals such as Beijing Medical Association, Family Medicine Clinic 15 hospitals and diabetes care network, county Health Inspection Center and nonprofit groups cross-industry integration of health and medical innovation services.Chen Jianren said that as the retinal area contains the retina and macular, in addition to the detection of complications of diabetes, The future also plans to promote equipment to the glasses line, so that the general public are more likely to maintain eye health.

Chen further said that in the future, the technology can be used in the field of wound care.With artificial intelligence to identify the proportion of wound tissue and image course changes, to help nurses to accurately understand the wound changes, and then through the cloud platform online, obtained the first time Professional wound care of human support, improve medical efficiency, will lead the development of smart medical new software industry and medical imaging equipment manufacturers value-added business opportunities.