The AI diagnostic technology from Alibaba DAMO Academy can accurately interpret CT images of suspected cases of new coronavirus within 20 seconds with an accuracy of the analysis results reaching 96%, Alibaba said on Saturday.
Nucleic acid testing is widely recognized as the main reference standard for the diagnosis of new coronavirus pneumonia as evidence of pathogenicity. With the accumulation of clinical diagnostic data, the characteristics of imaging big data of new coronavirus pneumonia gradually become clearer, and CT imaging diagnosis results become more and more important.
According to the fifth edition of the diagnosis and treatment plan published by the National Health Commission, clinical diagnosis does not need to rely on nucleic acid test results, and the clinical diagnosis results of CT images can be used as the criterion for judging cases of new coronavirus pneumonia.
The features of CT chest radiographs of patients with new type of coronavirus pneumonia are multiple changes in single or double lung, patchy or segmental ground-glass density, and other subtle changes.
The CT images of a patient with a new type of coronavirus pneumonia are about 300, which puts great pressure on the doctor's clinical diagnosis. The doctor's visual analysis of a CT image of a case takes about 5-15 minutes.
The DAMO Academy medical AI team has broken through the limitation of insufficient training data based on the latest diagnosis and treatment programs and papers published by multiple authoritative teams on the clinical characteristics of patients with new type of coronavirus pneumonia. Based on the CT image sample data of more than 5000 cases, it learns training For the sample lesion texture, a new AI algorithm model was developed.
Through NLP natural language processing of retrospective data and training of CT image recognition networks using CNN convolutional neural networks, AI can quickly identify the difference between new coronavirus pneumonia images and ordinary viral pneumonia images, and the final recognition accuracy rate is as high as 96%.
AI only takes less than 20 seconds to identify a case on average, greatly improving the diagnosis efficiency and reducing the pressure on doctors.
This is not the first time that DAMO Academy has demonstrated its technological power in the fight against epidemics.
Previously, on February 1st, the Zhejiang Center for Disease Control and Prevention launched an automated whole-genome detection and analysis platform, which was jointly developed by Zhejiang Province Center for Disease Control and Prevention with DAMO Academy and Jieyi Biological.
With the aid of the AI algorithm developed by Alibaba DAMO Academy, the original hours of genetic analysis of suspected cases has been shortened to half an hour, and Jieyi's new detection equipment has greatly shortened the diagnosis time of suspected patients.
According to the DAMO Academy, this platform can also effectively prevent missed detection of virus mutations and accurately detect possible virus mutations, providing support for subsequent vaccine and drug development.