Amidst the ongoing global COVID-19 pandemic, new virus variants have emerged with characteristics such as short incubation periods, rapid transmission rates, and high viral loads. Comprehensive, rapid, and accurate detection holds significant importance for both China and global pandemic prevention and control efforts.

The research team, led by Professor Huimin Zhao and Professor Jinchang Ren, has focused on the segmentation processing of chest X-ray superpixel images for COVID-19 and intelligent computing in the AI spatial domain. They have developed the SC2Net (Segmentation-based COVID-19 classification network), effectively addressing the precise diagnosis of early COVID-19 infections.

SC2Net comprises the CLSeg network for lung image segmentation and the SANet (Spatial Attention Network). Experiments conducted on the Spanish COVIDGR 1.0 dataset (Hospital Universitario Clínico San Cecilio, Granada, Spain) demonstrate that SC2Net achieves an average diagnostic accuracy of 84.23%, surpassing the accuracy of existing methods, such as FuCiT-Net and COVID-SDNet, by 3% to 4.8%. These research achievements have been published in authoritative journals, including IEEE Journal of Biomedical and Health Informatics (IF=7.021) and IEEE Transactions on Cybernetics (IF=19.118).

The team is currently actively collaborating with the Affiliated Hospital of Guangzhou Medical University, aiming to enhance diagnostic efficacy through clinical applications further.