ICIG 2021 Symposium:Medical Imaging Computing: From Theory to Applications

Date&Time:13:30-15:30, December 27, 2021   Location: Duke Ballroom

IntroMedical Image Computing (MIC) pioneers research in machine learning and information processing, with the particular aim of improving patient care by systematic image data analytics. Many works have been developed to structure and quantify imaging information from multiple time-points and imaging technologies, e.g. magnetic resonance imaging or computer tomography, and link it with clinical and biological parameters. As an initiator and co-coordinator of China Society of Image and Graphics, we organize this symposium to review the developments at the core of computer science, with applications in but also beyond medicine.


Organizer: Huafeng Liu (Zhejiang University)




Automatic diagnosis of cardiovascular interventional image

Chenxi Huang


Unsupervised deep learning based denoising for PET image

Jianan Cui


Computational Diffusion MRI and Its Applications in Schizophrenia

Sangma Xie


Artificial Intelligence in Cardiac Image Analysis

Zhifan Gao


Title: Automatic diagnosis of cardiovascular interventional image

Speaker: Chenxi Huang,Xiamen University

AbstractWith the accelerating aging of the population in China, the prevalence and mortality of cardiovascular diseases have increased significantly. Cardiovascular disease is the leading cause of death among urban and rural residents, which is higher than other diseases such as tumors and respiratory systems. Traditional cardiovascular images diagnosis mainly relies on manual subjective analysis, which requires a lot of time and cost. At the same time, due to the influence of factors like intravascular ultrasound imaging mechanism and equipment, the generated image has a unique speckle noise, which seriously affects the image quality. At present, intravascular ultrasound (IVUS) and optical coherence tomography (OCT) are the two main imaging methods used in medical practice. Therefore, we mainly focused on IVUS and OCT images to study the denoising algorithm of cardiovascular images to obtain clear images. Secondly, we use deep learning methods to segment the vascular membrane structure, stents, plaques, etc., and finally study the blood vessel three-dimensional visualization. The research results can not only assist doctors in diagnosis and treatment, but also improve doctors’ work efficiency and detection accuracy. Besides, this research can deepen researchers’ understanding of the mechanism of cardiovascular disease and provide more feasibility for the research and development of cardiovascular disease.

Biography: Dr. Chenxi Huang is currently an Assistant Professor with the School of Informatics, Xiamen University. His research interests include image processing, image reconstruction, data fusion, 3D visualization, and machine learning. He serves as an Associate Editor for Journal of Medical Imaging and Health Informatics (SCIE) and Frontiers in Medical Technology. He is also a reviewer for IEEE Access, Neurocomputing, Peerj, Journal of Grid computing, IEEE journal of biomedical imaging and health informatics, IEEE Transactions on Emerging Topics in Computational Intelligence, Journal of Medical Imaging and Health Informatics, and other SCI journals. He has published many high-level academic papers in related research fields. Recently, he has published 20+ SCI journal papers as the first author or corresponding author in ACM Transactions on Multimedia Computing, Communications, and Applications, IEEE Transactions on Instrumentation and Measurement, Complexity and Frontiers in Neuroscience.


Title: Unsupervised deep learning based denoising for PET image

Speaker: Jianan CuiZhejiang University

Abstract: Positron emission tomography (PET) is a powerful imaging method in nuclear medicine that can detect the metabolic activity of molecules in living tissues. PET image generally has the problem of low signal-to-noise ratio, which will influence the accuracy of diagnosis. Recently, deep learning has been widely used in the field of medical imaging denoising but usually, deep learning denoising methods need the use of low-quality/high-quality image pairs to supervise the network training. However, in real clinical practice, it is difficult to obtain high-quality PET images for training. Therefore, we proposed an unsupervised deep learning denoising framework, Conditional Deep Image Prior, for PET image denoising. Our experiments showed that based on this framework, excellent denoising effects can be achieved only using the noisy PET image and the registered CT/MR image.

Biography: Jianan Cui is a postdoctoral fellow at Zhejiang University. She received her B.E degree from Zhejiang University in 2015 and her Ph.D. degree in information sensing and instrumentation from Zhejiang University in 2020. During her Ph.D., She was also joint trained at Massachusetts General Hospital/Harvard Medical School for two years and a half. Her research focuses on PET image reconstruction, PET image denoising, and ASL image super resolution. She has published more than 10 papers in international conferences and journals. And she received the Women in Imaging Award in Fully3D 2019.


Title: Computational Diffusion MRI and Its Applications in Schizophrenia

Speaker: Sangma Xie,Hangzhou Dianzi University

Abstract: White matter tracts consisting of axons that connect distinct brain regions are important for brain functions. Diffusion MRI (dMRI), together with computational tractography, provide non-invasive methods to reveal the microstructure of the white matter of the in-vivo human brain. The first part of this talk will provide a brief methodological overview of dMRI analysis and tractography. DiffusionKit, a light, one-stop, cross-platform solution for dMRI data analysis, will be presented. Multiple lines of evidence indicate that the microstructure of white matter tracts is implicated in the pathophysiology of schizophrenia. In the part of applications in schizophrenia, we investigated the language pathways in schizophrenia patients with auditory verbal hallucination (AVH). Reproducible abnormal pattern of arcuate fasciculus in schizophrenia with AVH base on multi-site diffusion MRI data were found. Besides, we explored the 25 major white matter tracts with dMRI-based tractography in schizophrenia patients and their unaffected first-degree relatives to investigate disease-related and/or genetic risk-related anatomical variations in schizophrenia. Our results suggest that alterations in the left superior longitudinal fasciculus and anterior thalamic radiation are features of established illness, which may serve as neural substrates for the psychopathology of schizophrenia.

Biography: Sangma Xie received the B.E. degree in computer science and technology from Lanzhou University in 2011 and Ph.D. degree in pattern recognition and intelligent system from Institute of Automation, Chinese Academy of Sciences in 2017. From 2017.07 to now, he works as Lecturer at Hangzhou Dianzi University. His current research interests include computational methods of dMRI, software for dMRI analysis, and brain network researches of neuropsychiatric diseases using dMRI. He has published more than 10 papers in international conferences and journals, and developed DiffusionKit toolbox. He has presided over projects of National Nature Science Foundation of China and Zhejiang Provincial Natural Science Foundation of China.


Title: Artificial Intelligence in Cardiac Image Analysis

Speaker: Zhifan Gao,Sun Yat-sen University

Abstract: According to World Health Organization, cardiac disease has been the major reason leading to the mortality in all causes. The Report on Cardiovascular Disease in China shows that the number of patients with cardiac disease in China has reached 330 million. The artificial intelligence has shown the benefit to the medical intervention, and gradually indicated its effectiveness by combining with the cardiac medical image. This talk introduces recent studies proposed by our research group in the medical image analysis and artificial intelligence for cardiac disease. It focuses on two aspects, the morphological abnormality and the functional abnormality, and then introduce several cardiac image analysis tasks in various image modalities. This talk further explains how these studies can decrease the labor consumption, improve efficiency, and reduce costs and risks during the diagnosis and treatment.

Biography: Dr. Zhifan Gao is currently an associate professor with School of Biomedical Engineering, Sun Yat-sen University, China. Before that, he was a postdoctoral fellow with Western University, Canada from 2018 to 2020. His research interests include medical image analysis and machine learning. He has published 40+ peer reviewed papers (such as IEEE TMI, IEEE TNNLS, Medical Image Analysis), including four ESI highly cited papers and one ESI hot paper. He serves as the program committee member for IJCAI 2021 and AAAI 2021. He won CAAI Wuwenjun Outstanding Young Scientist 2020, President Award of Chinese Academy of Sciences 2017, and Publons Top Peer Reviewer Award 2019.