Development and Evaluation of an AI System for COVID-19

Cheng Jin; Weixiang Chen; Yukun Cao; Zhanwei Xu; Xin Zhang; Lei Deng; Chuansheng Zheng; Jie Zhou; Heshui Shi; Jianjiang Feng

It is important to diagnosis COVID-19 as quick and accurate as possible for controlling the spread of the disease and treating patients. Even though reverse transcription-polymerase chain reaction (RT-PCR) is still ground truth of COVID-19 diagnosis, the sensitivity of RT-PCR is not high enough for low viral load present in test specimens or laboratory error [2] , and kits of RT-PCR are in short of supply in many areas [3] . By contrast, chest CT can show early lesions in the lung and, if diagnosed by experienced radiologists, can achieve high sensitivity [4] . In addition, CT scans are widely available and economic. At present, the diagnosis of chest CT depends on the radiologists, which has some problems. Firstly, chest CT contains hundreds of slices, which takes a long time to diagnose. Secondly, COVID-19, as a new lung disease, has similar manifestations with various types of pneumonia4. Radiologists need to accumulate a lot of CT diagnostic experience to achieve a high diagnostic performance. In some outbreak areas, many suspected patients are in urgent need of diagnosis and proper treatment, and many CT scans are performed every day. There is an urgent shortage of radiologists with high diagnosis performance for COVID- 19 .