In response to the outbreak of coronavirus in China, the Chinese Academy of Sciences (CAS) has launched a priority program for the containment and prevention of Covid-19. In addition, CAS has formed a joint task force from the CAS Clinical Research Hospital for using scientific studies and technologies to prevent and control the epidemic (hereinafter referred to as the ‘CAS Headquarters for Tackling Key Scientific and Technological Problems’ or ‘CASHTST’). As a company with extensive knowledge, well-developed product applications and hands-on experience in medical imaging recognition and auxiliary diagnosis, iFLYTEK is well-suited to play a part and quickly assumed the task of building a new A.I. image-assisted diagnosis platform for Covid-19.
With the coordination of CASHTST, the iFLYTEK task team was activated and deployed the system in just three days. Team members were stationed on site to make sure the system was adjusted on a daily basis through data iteration and algorithm optimization.
At present, the A.I. image-assisted diagnosis system for Covid-19 completes in 3 seconds, with 100% positive cases recalled, and 90% of legions recalled, providing doctors with accurate and efficient diagnostic information.
4D contrast analysis for accurate evaluation of lung lesions and their changes in patients with Covid-19.
During the epidemic, imaging doctors must make fast and accurate readings of chest CT results from confirmed patients, suspected patients, and outpatients with fever. They are physically and mentally challenged to the limits and constantly under huge stress.
Traditional imaging technologies only offer basic reviews and certain measurements in length, but are not able to recognize, segment, or measure the size and density of the lesions. When the doctors want to compare patients’ historical images and make further assessment on the changes of the lesions, traditional tools that can only manage to retrieve the data still left doctors in a grey area for making final calls.
Compared with traditional imaging tools, the A.I. image-assisted diagnosis platform for Covid-19 provides 4D contrast and analysis which quickly identifies the number of lesions and changes in their volume and density out of massive data collected from multiple periods, giving imaging doctors first hand development of the patients’ conditions, which greatly improves the accuracy in diagnosis and treatment.
For patients diagnosed with Covid-19 pneumonia, their CT images show increased lesions and complicated development, which were not seen before. By using the innovative 4D contrast analysis function in pairing the before-and-after images, and by using the various auxiliary tools, doctors are able to quickly determine changes in the patient’s conditions.
After the implementation and uses of the A.I. image-assisted diagnostic platform for Covid-19, imaging physicians at frontline hospitals and treatment centers highly approved its 4D contrast and analytic capabilities. They agreed that the quantitative comparison of lesions and density can quickly lead to the determination of the changes for lung lesions in patients with infections, and help the imaging professionals in making correct decisions.
In practice, it takes imaging physicians about 5 to 15 minutes to diagnose each patient by analyzing hundreds of CT slices by layers. For patients suspected with coronavirus pneumonia, their historical images also need to be carefully reviewed, which doubled the workload. iFLYTEK’s A.I. medical imaging system, designed from the real needs of imaging physicians, can greatly improve their work efficiency, and effectively reduce missed diagnosis and misdiagnosis.
Multi-mode auxiliary diagnosis and accurate identification of novel coronavirus pneumonia.
For suspected outpatients with fever and other symptoms, it is necessary for imaging physicians to be able to quickly complete the screening for signs of the virus, and issue early alerts. The iFLYTEK system integrates medical records, CT images, lab examination results and other critical information for a multi-mode diagnosis, improving the accuracy of CT images interpretation, reducing the rate of misdiagnosis, shortening hospital wait times, and eventually lowering the risk of cross-infection.
When the CAS special task forces were looking for working partners to build the A.I. medical imaging assisted diagnosis system, iFLYTEK was among their top choices, attributed to the company’s long-term commitment to researching and developing artificial intelligence for communities to achieve better medical capabilities.
· In August 2017, iFLYTEK won first place in LUNA, an international authoritative evaluation in the field of international medical imaging, with significant advantages, and set a new world record.
· In March 2018, iFLYTEK won first place in the most difficult MA segmentation task of IDRiD fundus image analysis competition held by ISBI, the top international medical imaging conference.
· In July 2019, iFLYTEK broke the world record of CT liver segmentation in CHAOS, an international authoritative evaluation of medical imaging.
· iFLYTEK passed the written test of national clinical practitioner examination and received a high score of 456. iFLYTEK A.I. Medical Assistant has been applied in many healthcare providers for auxiliary diagnosis.
The successful application of A.I. medical imaging assisted diagnosis system proves the importance of using artificial intelligence and other technologies in containing and curing the viral disease. Aiming to help more people, iFLYTEK teams have worked closely with colleagues from the First Affiliated Hospital of University of Science and Technology of China, in providing A.I. assisted medical imaging services to more than 1,200 medical facilities on their jointly developed A.I. platform. iFLYTEK A.I. Medical Assistants have also been used extensively by 12,000 primary healthcare providers and 36,000 primary medical practitioners, as well as by epidemic prevention and control headquarters, health committees and hospitals in 30 provinces and cities including Hubei, Beijing, Shanghai, Zhejiang and Anhui.