New version of AI for medical imaging sent to Beijing YouAn Hospital

Beijing-headquartered Huiying Medical Technology Co Ltd, a high-tech enterprise focusing on artificial intelligence for medical imaging, said it recently provided a new version of medical AI system to Beijing YouAn Hospital to assist in the fight against the COVID-19 epidemic.

Attached to Capital University of Medical Science, the hospital is one of China's best for treating infectious disease, and many of its medical professionals are on the front line fighting against the COVID-19.

To lend a helping hand to medical workers who are overloaded with work diagnosing and treating the COVID-19 patient, the company collected medical data on more than 1,000 confirmed COVID-19 patients as quickly as possible, which was then labeled by professional doctors -- defining normal elements and categorizing abnormal ones.

Combined with the company's existing database from more than 20,000 pneumonia cases, the newly labeled data on COVID-19 cases was then processed with deep learning and transfer learning technologies to create accurate modeling on a full-data platform. This means the AI system is able to realize self-iteration and model optimization to automatically adapt to medical images with different layers of thicknesses from different hospitals and different equipment.

As a result, its detection rate and accuracy rate on pneumonia lesions reach 96 percent and can deliver a diagnosis within 2-3 seconds with more than 500 computer tomography scans when combined with technologies, such as efficient processors and lightweight network models, to improve the efficiency of the algorithm.

The AI system will assist doctors in making a diagnosis, which is of great help in clinical practice, based on its ability to accurately position and distinguish lesions.

During diagnosis and treatment, the system can provide quantitative data on lesion location, size, change in size and degrees of illness severity, among many others, to support efficient and accurate clinical assessment of disease progress and treatment efficacy.

Meanwhile, combined with the latest version of the diagnosis and treatment plan for the COVID-19, the AI system can provide interactive and structured reports on CT examinations of the novel coronavirus pneumonia. It also can help make medical image reports structured, intelligent and standardized to improve a hospital's information construction and the quality of medical image reports, and assist the collection of single disease data to establish an effective database for further mining of the data in the future.

Li Hongjun, director of Beijing YouAn Hospital's Radiology Department, said after using the AI system that it has reached a new level in speed and accuracy of diagnosis.

The system clearly shows the location and data of lesions, and its automatic and quantitative comparison function also clearly shows changes in patients' conditions, which is of great help for follow-up treatment in the front line and epidemic prevention and control, Li said.

The company said it will continue its efforts to help medical workers during the prevention and control of the epidemic, as well as in the future for everyday treatment of patients.