With the new coronavirus epidemic, physicians need to seek a more rapid diagnostic approach to identify suspected cases. In addition to reagent testing, chest CT image identification has become the main means of screening patients. In early February of this year, the National Health Commission released the “New Coronavirus Infection Pneumonia Diagnostic and Treatment Means (Trial Fifth Edition)”, and CT imaging results were included in the clinical diagnosis criteria in Hubei Province, “suspected cases with pneumonia imaging features”, which is a clinical diagnosis case. Once the fifth edition of the trial treatment plan was released, the workload of CT scans in hospitals in all provinces of China increased dramatically. In actual clinical practice, each patient may need to review CT once in 2~4 days, and about 300 pictures have to be taken in one CT, together with suspected patients and close contacts screening, the massive amount of new CT images generated every day is a huge work pressure for front-line imaging doctors. AI assisted to improve the CT image reading rate Previously, the hospital-wide CT volume of some of Wuhan’s new crown pneumonia designated hospitals had exceeded 1,000 cases a day. If we count 300 photos for one CT case, it means that a hospital imaging doctor has to read 300,000 photos a day. In clinical diagnosis, doctors are less efficient in identifying CT images with the naked eye, taking 5-15 minutes for each case. This is where the artificial intelligence technology of technology companies comes into play. We see a lot of demand for CT in major hospitals,” said Shiyuan Liu, director of the imaging department at Shanghai Changzheng Hospital and chairman of the China Medical Imaging AI Industry-Academia-Research Alliance. It is used for AI to assist radiologists for early identification of lesions, determination of lesion changes through deep learning, and quantitative analysis is a much-needed technology in the industry”. To date, GE Healthcare, Shang Tang Technology, Alibaba, Huawei and other companies have used AI medical imaging systems to fight the new coronavirus outbreak. On March 5, GE Healthcare launched “Smart Win New Coronavirus LK2.0,” a CT image intelligence analysis platform technology for New Coronavirus pneumonia, for scientific research to assist in the accurate analysis of early, suspected New Coronavirus pneumonia. GE Healthcare said, “Smart Win New Crown LK2.0” is a software platform based on the principle of image genomics, combined with artificial intelligence and image processing and other technologies, to conduct intelligent analysis of new crown CT images. On Feb. 21, ShangTong Technology upgraded its self-developed “Sense-Lung” lung AI intelligent analysis product with the new crown pneumonia function, and connected it to the medical image reading platform of its partners in the mode of cloud service, helping more than ten front-line hospitals and many local medical institutions through remote ways. In addition to the above two, Alibaba’s “New Crown Pneumonia” product has been upgraded with a cloud service model to access the medical image reading platform of its partners. In addition to the above two, Alibaba’s Dharma Institute and Ali Cloud, Huawei Cloud and Huazhong University of Science and Technology, all developed and launched the new coronavirus pneumonia AI-assisted medical image quantitative analysis service. From the technical level, the introduction of AI for CT image reading can help doctors improve efficiency and provide valuable reference for doctors who are not yet experienced, thus making the whole diagnosis faster. However, industry experts say that the integrated application of multiple tools should be advocated, because each one is difficult to achieve 100% accuracy, and doctors should have their own set of logic to use. As for the clinical application of diagnosis, doctors will also consider the symptoms, the main complaints and past medical history and other factors. Nucleic acid test results have been questioned Previously, an article titled “First-line doctors cry out: replace nucleic acid test with CT as soon as possible as the standard for confirming the diagnosis of new coronavirus pneumonia” received widespread attention, in which Zhang Xiaochun, deputy director of the imaging department of Wuhan University Central South Hospital, was quoted as crying out in his circle of friends: stop believing in nucleic acid test and strongly recommend CT images as the main basis for the current 2019nCoV (diagnosis). In fact, it is a diagnostic imaging method. In fact, CT imaging, which is a diagnostic imaging method, and nucleic acid testing, which is pathogenic evidence, are two different dimensions to look at a thing. In an interview with China Science News, Gu Bing, a professor at Xuzhou Medical University, said that CT cannot distinguish which virus a patient is infected with, while nucleic acid testing can bulk screen it out, and that pathogenic evidence plays an important role in infectious diseases. Only when imaging and laboratory tests go hand in hand, it is more conducive to diagnosis and better prevention and control of the epidemic. The PCR kit for New Coronavirus detection works roughly as follows: by extracting RNA from a patient sample, performing reverse transcription polymerase chain reaction (RT-PCR), amplifying the trace amount of viral information in the sample through amplification reactions, and finally reading the signal as fluorescence. If the signal is positive after PCR, then it is safe to say that the virus is present in the sample, and vice versa, indicating that there is no infection. At the beginning of the epidemic, many manufacturers put their reagents on the market as soon as they were developed, without doing enough performance validation tests, and there was uneven quality. As a result, the nucleic acid test results also showed bias. False positives/false negatives are not only related to the quality of the kit, but also to the timely and correct collection of specimens. False negatives can be caused if specimens are collected incorrectly, if specimens are stored for too long (viral nucleic acid degradation), etc. CT imaging and nucleic acid testing should complement each other Screening for new coronavirus infection, choose chest CT or nucleic acid testing? A recent retrospective study of 1014 infected patients at Wuhan Tongji Hospital found that nucleic acid testing has poor sensitivity, good specificity, and is easy to miss; chest CT has good sensitivity, poor specificity, and is easy to misdiagnose. Why is there some variability in the data results of CT test and nucleic acid assay? Xu Haibo, director of the Department of Medical Imaging at Wuhan University Central South Hospital, said, “These are two different standards used by the clinic. The positive result of viral nucleic acid test is pathological or pathogenic as the testing standard. In the actual clinical process, 80% or 90% of the cases should be diagnosed in this way. Diagnosing a disease never requires a comprehensive analysis by physicians, especially for the majority of primary care hospitals, which rely solely on imaging or pathogenic evidence to detect new coronavirus pneumonia, is very difficult. In Wuhan, for example, because there are more patients in the early stage, nucleic acid testing alone is prone to miss the diagnosis, and CT screening is more advantageous for early detection of patients and early isolation. In areas outside Hubei, such as Qinghai, Gansu, Ningxia and other areas with fewer patients, the appropriateness of routine chest CT screening in outpatient clinics needs to be debated and further studied. Experts believe that the nuclear test has low sensitivity but no radiation and good specificity, so it may be more appropriate if it can be popularized. Zhang Xiaochun later in a media interview also said that the release of the article is not to deny the results of the nucleic acid test, but that it is the ultimate means of detection, in the beginning by the yield, sampling methods, in Wuhan can not achieve the effect of cutting off the source of infection prevention and control. However, he believes that this recommendation is only suitable for disaster areas, not for ordinary, scattered case screening. Ordinary screening of disseminated diseases should be done for differential diagnosis, which CT cannot do. Content source: Observer.com