AI-assisted clinical diagnosis technology can effectively improve the efficiency and accuracy of clinical diagnosis, and with the development of deep learning, research on deep neural network models for AI-assisted clinical diagnosis is booming. However, AI-assisted clinical diagnosis faces some challenges, including the inherent characteristics of medical imaging modalities that may lead to low quality of medical images, the camera environment may introduce interference noise to medical images and videos, and the lack of accurate expert annotation data, etc. This lecture will focus on solving the difficult problems of AI-assisted clinical diagnosis, and introduce the relevant research progress of our research group. First, to solve the problems of low image quality and scarcity of accurate expert annotation data, a framework of deep learning rough segmentation combined with automatic evolution of image processing algorithms was proposed. Next, to solve the problem that there are many distracting objects in the endoscopic field of view in the minimally invasive spine decompression surgery environment, a video inter-frame attention and channel self-attention module is proposed to obtain global feature information, and the spinal nerve area was segmented from the endoscopic video images. Then, aiming at intelligent cervical spine function assessment, a lightweight neural network model analyzing the symptoms of abnormal hand movements was proposed, which is applied in clinical practice and effectively assists clinical diagnosis. Finally, the research progress of our research group on the design and training of deep neural network models is introduced, which can speed up model training and inference while improving the accuracy of computer vision tasks. 

Biography:

Dr. Junying Chen received the B.E. degree in electronic and information engineering from Zhejiang University in 2007, and the Ph.D. degree in electrical and electronic engineering from The University of Hong Kong in 2013. She is currently an Associate Professor with the School of Software Engineering, South China University of Technology. She is the Assistant Director of the Key Laboratory of Big Data and Intelligent Robot (KLBDIR) of the Ministry of Education, and the Principal Investigator (PI) of the Intelligent Medical Image Processing Laboratory of KLBDIR. She is the Distinguished Speaker of ACM, the Distinguished Member of China Computer Federation, and the Senior Members of IEEE, China Society of Image and Graphics, and Chinese Society of Biomedical Engineering. Her research interests include medical image processing, AI-assisted clinical diagnosis, deep neural network models, multi-source/multi-modal feature fusion, high-performance computing, etc. She is the PI of research projects funded by National Natural Science Foundation of China, Guangdong Natural Science Foundation, and Guangzhou Science and Technology Program. She published more than 40 academic papers as the first author or corresponding author in IEEE TPAMI, TNNLS, TMI, etc., is authorized 3 invention patents, 5 utility model patents, and 27 software copyrights, published 2 books, edited 3 conference proceedings, and received the Second Prize for Technological Invention from China Computer Federation Science and Technology Award and Second Prizes of Excellent Papers of Guangdong Computer Academy. The research results have been promoted and applied in many hospitals in Guangdong Province, Jiangxi Province, etc., and have been selected into "Technical Public Welfare Case Collection in China Computer Federation 2022".