Recently, China National Medical Products Administration approved shenzhen Silicon Intelligent Technology Co., Ltd. innovative products “diabetic retinopathy fundus image auxiliary diagnosis software” and Shanghai Eagle Tong Medical Technology Co., Ltd. innovative products “diabetic retinopathy fundus image auxiliary diagnosis software”. This is the first time that Chinese AI technology has been applied to image-assisted diagnosis in ophthalmology. At present, there is no domestic approval of other similar products on the market.
Provide more fundus screening for diabetes patients
According to the Global Diabetes Overview (Ninth Edition) released by the International Diabetes Federation in November 2019, the total number of diabetes patients in China was about 116.4 million as of 2019. Diabetic retinopathy (DR) is the most common complication of diabetes. According to epidemiological survey results, about 1/3 of diabetic patients have signs of DR, and 1/3 of them may have retinopathy that threatens vision. DR focuses on early prevention, early detection and early intervention.
The screening is mainly carried out by fundus photography and doctors’ film reading. However, clinical medical practices such as disease evaluation, follow-up and prognosis evaluation of DR patients all require a lot of medical resources. However, the number of professional ophthalmologists in China is insufficient and the distribution is uneven, and the vast majority of diabetic patients fail to receive retinal examination in time. As an independent artificial intelligence software, DR fundus image auxiliary diagnosis software can provide medical services for clinical auxiliary diagnosis and improve the chance for diabetes patients to receive DR screening.
The approved products of Shenzhen Silica-based Intelligent Technology Co., Ltd. are composed of usb disk, and their functional modules include DR assisted diagnosis cloud, doctor diagnosis client and system management terminal. It is applicable to analyze binocular color fundus images of adult diabetes patients, and provide auxiliary diagnostic suggestions for medical practitioners on whether stage II or above DR can be seen and further medical examination.
Shanghai Airdoc medical Technology Co., LTD. (Airdoc) products include client software and server software. The server is deployed on Ali Cloud, and the client is deployed on the client computer. The server and the client exchange encrypted data through the Internet. It is suitable for the analysis of binocular non-dilated color fundus images of adult diabetic patients, and provides auxiliary diagnostic suggestions for medical practitioners on whether moderate non-proliferative diabetic retinopathy (including diabetic retinopathy) is found and further medical examination.
It is understood that the two approved products both use the self-designed network structure based on convolutional neural network, and train and verify the algorithm model through classified annotated fundus image data. The product obtains color fundus photos of patients and uses deep learning algorithm to calculate and analyze the images to obtain auxiliary diagnosis suggestions for diabetic retinopathy, which can be provided to qualified clinicians as a reference.
More AI products will benefit the clinic
In recent years, the innovative research and development of medical AI products ushered in the development boom, and artificial intelligence has been widely applied in DR disease diagnosis and lesion classification.
In April 2018, the US Food and Drug Administration (FDA) announced on its official website that it approved idX-DR, the world’s first commercially available AUTONOMOUS DR screening AI product based on CNN deep learning, for primary care institutions. The device automatically diagnoses DR staging without the assistance of a doctor. The two approved products are the first batch of ai-assisted diagnostic software for fundus approved for market in China.
It is worth noting that in recent years, in order to accelerate the review and approval progress of artificial intelligence medical products, the State Food and Drug Administration has opened up a green channel. Since April 2019, keya Medical, Yingtong Medical, Shukun Technology, silicon-based intelligent and other DR and coronary artery products have entered the “green channel” of innovative medical devices. Some industry insiders said that at present, AI medical imaging application is the most mature, and it is the only field with approved products. Most of the applications start with a single disease, mainly with simple image training. From the perspective of application, fundus screening is one of the disease fields with the most enterprise layout.
In addition to ophthalmic AI products, other AI products in China also ushered in breakthroughs. On January 15 this year, the State Food and Drug Administration announced that the first domestic artificial intelligence medical device — Beijing Kunlun Medical Cloud Technology Co., LTD. ‘s innovative product coronary blood flow reserve score calculation software has been approved for market. In February, THE ARTIFICIAL intelligence “ECG Platform” (AI-ECG Platform) independently developed by Shenzhen Kaiver Electronics Co., LTD., a wholly-owned subsidiary of THE AI Division of LEp (Beijing) Medical Device Co., LTD., was approved.
At present, the Medical device Technical Evaluation Center of sFDA has led and cooperated with 14 government-industry-university-research and application entities in the field of medical devices to form 12 cross-department, cross-field and cross-industry working groups to accelerate the transformation and application of artificial intelligence scientific and technological achievements in the field of medical devices.
In July 2019, the research group of artificial intelligence Medical device Safety and effectiveness Evaluation research Project released the world’s first technical review points on artificial intelligence medical device “Deep Learning Decision-assisted Medical Device Software Review Points”, promoting the development of artificial intelligence industry.