Source: Ophthalmic Engineering Research Center, ZOC

Written by: Peng Xiao, Ophthalmic Engineering Research Center, ZOC

Proofread by: Jiawei Wang

Edited by: Xianjing Wei

On August 7, 2020, "Diabetic Retinopathy Fundus Image Auxiliary Diagnosis Software" (DR Aided Diagnosis Software), a medical imaging artificial intelligence auxiliary diagnosis product with independent intellectual property rights, was approved by the State Food and Drug Administration for the third category of medical device certification, becoming the first fundus diagnosis software based on artificial intelligence technology approved in China. The project was developed by the national key research and development plan "Digital Diagnostic and Treatment Equipment", the key special item "Development and Application of Ophthalmic Multimodal Imaging and Artificial Intelligence Diagnosis and Treatment System" (2017YFC0112400). The project is led by Professor Yuan Jin, the project chief, and conducted by Zhongshan Ophthalmic Center of Sun Yat-sen University as the leading unit, and Shenzhen Silicon Intelligent Technology Co., Ltd. as the participating unit.

The National Key R&D Project successfully passed the performance acceptance of the Ministry of Science and Technology

Based on more than 103,000 Chinese fundus images and cross-labeling by dozens of fundus doctors, the product uses a self-designed convolutional neural network structure to train and verify the model for diabetic retinopathy. The completion of a multi-center prospective clinical trial of 1,000 samples was led by Zhongshan Ophthalmic Center and participation of Peking University People’s Hospital and Affiliated Eye Optometry Hospital of Wenzhou Medical University. The key performance indicators are sensitivity of 87.29% and specificity of 95.51%. The performance of the product advanced in global research. Related core technologies have been declared for 9 domestic invention patents and 2 international invention patents (PCT), 4 of which have been authorized for invention patents.

Based on fundus images to assist in the diagnosis of diabetic retinopathy, the sensitivity is 87.29% and the specificity is 95.51%

DR diagnostic software obtains abundant professional experience of fundus specialists through deep learning and is widely used in endocrinology departments and primary hospitals over the 5G network to assist endocrinologists and primary general practitioners in the diagnosis and management of ophthalmic complications of diabetes. This can effectively help diabetic patients to get efficient diagnose and timely treatment, avoid the risk of blindness, promote tiered diagnosis and treatment, improve the medical capacity of primary hospitals, and enable high-end medical care to benefit the grassroots. After approval, products will be firstly applied in Hainan, Shandong province, and Shenzhen city.

5G ophthalmology AI cloud platform has covered primary eye hospitals in Zhaoqing, Guangdong Province