Medical Image Computing and Computer Assisted Intervention (MICCAI), the international Conference on medical imaging, concluded in Singapore on Sept 22, 2022. In the ultra-low dose PET(UDPET) Imaging challenge, Gaoyu Chen, Sheng Liu, Li Lv, Wenxiang Ding and other students from the School of Biomedical Engineering of Shanghai Jiao Tong University, led by Professor Qiu Huang, stood out from teams from famous universities and scientific research institutions in the world and won the champion of the competition.

This challenge is the first competition in the field of UDPET imaging, and is aimed to restore high-quality standard-dose images from low-dose reconstructed images provided by the sponsor from uExplorer and Siemens Biograph Vision Quadra systems. The down-sampling magnification of the detection data for low-dose images varies from 4 to 100 times. The lack of detection data results in a large amount of noise in low-dose images, which poses a great challenge to the mission.

Professor Huang's team through years of research experience in PET imaging field, fully analyzed data sets and the design of a set of efficient data preprocessing and deep learning training strategy. In only two months, they conducted a large number of comparative experiments, comprehensively considered the advantages and disadvantages of each model in different evaluation indicators, put forward a complete solution, and finally got the first place.