“Integration, Exchange, and Explore”
On April 1, the Dushu Forum: 2023 Dushu Lake Conference on Medical Imaging Computing (hereinafter referred to as “the conference”) kicked off at SIAR. The conference was sponsored by SIAR and Suzhou Industrial Park Administrative Committee (SIPAC). The 2-day conference brought together experts and scholars in the field of medical imaging computing from universities, research institutes and companies across the country. Through lectures, seminars, poster exchanges and group activities, scholars discussed the latest progress and future development of medical imaging computing.
The conference was chaired by Professor Zhou Shaohua, Executive Dean of School of Biomedical Engineering, USTC, Director of Center for Medical Imaging, Robotics, Analytic Computing and Learning (MIRACLE), SIAR, USTC.
Chen Miao, member of the Party Working Committee and Deputy Director of SIPAC, made an opening speech. Chen Miao extended a warm welcome to the experts and scholars present, introduced the development and achievements of Suzhou Industrial Park and Dushu Lake Science and Education Innovation District, and expressed the best wishes for the conference.
Chu Jiaru, Secretary of the Party Committee and Executive Dean of SIAR, introduced to the guests the overall situation of SIAR, as well as the development and achievements of the biomedical engineering discipline in Suzhou in accordance with local conditions and the integration of industry, education and research.
The conference invited six experts, Chen Xilin, Gao Jiahong, Yu Jingyi, Liu Haiyan, Liang Huiying and Zheng Yefeng, to give lectures and share their research results and experiences.
Professor Chen Xilin introduced self-supervised pre-training and its application in surgical vessel and instrument segmentation, and showed some attempts at self-supervised pre-training on surgical-related images and the progress made in surgical vessel and instrument segmentation combined with abdominal surgery scenes.
Professor Gao Jiahong introduced the research and development of AI technology and its application in brain science, medical imaging and brain-computer interface. Professor Gao pointed out that AI technology has been widely used and developed in the fields of brain science, medical imaging, and brain-computer interface. Application and research in these fields can promote the development and innovation of AI technology, and provide powerful technical support and tools for future basic neuroscience and clinical application research.
Professor Yu Jingyi introduced the application of Neural Radiance Field (NeRF) in medical imaging. Professor Yu first introduced how to apply NeRF technology to the field of sparse view CT reconstruction, and how to integrate the physical model to eliminate metal artifacts and correct motion blur. In addition, Professor Yu’s team has achieved high-resolution reconstruction rendering and high compression rate compression technology, and designed a dedicated chip for real-time CT rendering.
Professor Liu Haiyan introduced de novo protein design based on artificial intelligence. Professor Liu discussed the particularity of protein design (compared to other AI application scenarios) from the perspective of understanding the biophysical laws of protein structure and function, and introduced the latest progress in de novo protein structure design using the diffusion probability model.
Director Liang Huiying delivered a lecture under the topic of “The Foundation for the Intelligent Transformation of Digital Rubik’s Cube Medical Institutions”. Director Liang focused on the exploration of multi-modal medical big data full life cycle products, technologies, and scenarios, and proposed a complete data-driven methodology for the intelligent transformation of medical institutions to clarify their intelligent vision, implement intelligent practices, and enhance intelligent capabilities.
Director Zheng Yefeng gave a lecture named “Intelligent Learning of Medical Imaging Artificial Intelligence Based on Imperfect Clinical Data”. Director Zheng pointed out that data collected from clinics usually have various flaws. In order to learn a robust intelligent image analysis system from these imperfect clinical data, Director Zheng’s team proposed small-sample learning techniques to solve the problem of a small amount of labeled data, adopted high-fidelity image style transfer technology to reduce the impact of cross-center data differences, and pioneered the use of multiple doctors’ original annotations for learning, making full use of the implicit information in inconsistent labels to improve the robustness of the algorithm.
The conference set up seminars where 23 scholars introduced their work, displayed the cutting-edge achievements, technology applications and practical experience in their research fields. The audience also showed their enthusiasm and spirit of inquiry, asking questions about the speaker's research direction, methods, technical details and practical applications.
In order to promote face-to-face communication among experts and students, on April 2, a poster exchange was held in the exhibition hall of Mingde Building, SIAR, USTC. The exhibitors designed exquisite posters with pictures and texts, vividly displaying the content and results of their research.
With the theme of “Integration, Exchange, and Explore”, the conference provided a platform that not only deepened mutual exchanges and cooperation, but also shared ideas for promoting the development of medical imaging computing technology. In this open, inclusive, and communicative academic environment, the research results of medical imaging computing will become more fruitful, bringing more benefits and progress to the medical field.
