VALSE Webinar 20240522-13期 总第348期 计算机辅助诊疗：过去，现在和未来 ...

报告嘉宾：周纵苇 (Johns Hopkins University)

报告题目：Body Maps: Towards 3D Atlas of Human Body

报告嘉宾：骆路阳 (香港科技大学)

报告题目：Understanding and Learning from Imperfect Medical Data

Panel嘉宾：

周纵苇 (Johns Hopkins University)、骆路阳 (香港科技大学)、韩楚 (广东省人民医院)、张建鹏 (阿里巴巴达摩院)、夏勇 (西北工业大学)

Panel议题：

1. 回顾计算机辅助诊疗的发展历史，现在面临的挑战和未来的机遇有哪些？

2. 面向医疗场景的计算机辅助诊疗系统开发具有重要临床意义，同时面临数据不完备的情况，包括数据规模相对较小，标注昂贵等，未来哪些途径有望解决这些问题？

3. 生成式学习在自然场景取得了巨大成功，在计算机辅助诊疗方面有哪些应用？

4. 深度大模型取得了飞速进展，多模态医疗大模型如何更好的赋能医疗？

*欢迎大家在下方留言提出主题相关问题，主持人和panel嘉宾会从中选择若干热度高的问题加入panel议题！

报告嘉宾：周纵苇 (Johns Hopkins University)

报告时间：2024年5月22日 (星期三)晚上20:00 (北京时间)

报告题目：Body Maps: Towards 3D Atlas of Human Body

报告人简介：

Zongwei Zhou is an assistant research scientist at Johns Hopkins University. His research focuses on developing novel methods to reduce the annotation efforts for computer-aided detection and diagnosis. Zongwei received the AMIA Doctoral Dissertation Award in 2022, the Elsevier-MedIA Best Paper Award in 2020, and the MICCAI Young Scientist Award in 2019. In addition to seven U.S. patents, Zongwei has published over 30 peer-reviewed journal/conference articles, two of which have been ranked among the most popular articles in IEEE TMI and the highest-cited article in EJNMMI Research. He was named the top 2% of Scientists released by Stanford University in 2022 and 2023.

个人主页：

https://www.zongweiz.com

报告摘要：

Cancer, a leading cause of mortality, can be effectively treated if detected in its early stages. However, early detection is challenging for both humans and computers. While AI can identify details beyond human perception, delineate anatomical structures, and localize abnormalities in medical images, the training of these algorithms requires large-scale datasets, comprehensive annotations. Several disciplines, including natural language processing (e.g., GPTs), representation learning (e.g., MAE), and image segmentation (e.g., SAMs), have witnessed the transformative power of scaling data for AI advancement, but this concept remains relatively underexplored in medical imaging due to the inherent challenges in data and annotation curation. This talk seeks to bridge this gap by focusing on datasets and annotations that are integral to the analysis of medical images, particularly for early cancer detection.

参考文献：

[1] Liu, Jie, Yixiao Zhang, Jie-Neng Chen, Junfei Xiao, Yongyi Lu, Bennett A Landman, Yixuan Yuan, Alan Yuille, Yucheng Tang, and Zongwei Zhou. "Clip-driven universal model for organ segmentation and tumor detection." ICCV (2023). 

https://www.cs.jhu.edu/~alanlab/Pubs23/liu2023clip.pdf

[2] Qu, Chongyu, Tiezheng Zhang, Hualin Qiao, Yucheng Tang, Alan L. Yuille, and Zongwei Zhou. "Abdomenatlas-8k: Annotating 8,000 CT volumes for multi-organ segmentation in three weeks." NeurIPS (2023). 

https://www.cs.jhu.edu/~alanlab/Pubs23/qu2023abdomenatlas.pdf

[3] Li, Wenxuan, Alan Yuille, and Zongwei Zhou. "How well do supervised models transfer to 3d image segmentation?" ICLR (2024). 

https://www.cs.jhu.edu/~alanlab/Pubs23/li2023suprem.pdf

[4] Chen, Qi, Xiaoxi Chen, Haorui Song, Zhiwei Xiong, Alan Yuille, Chen Wei, and Zongwei Zhou. "Towards Generalizable Tumor Synthesis." CVPR (2024). 

https://www.cs.jhu.edu/~alanlab/Pubs24/chen2024towards.pdf

[5] Xiang, Tiange, Yixiao Zhang, Yongyi Lu, Alan Yuille, Chaoyi Zhang, Weidong Cai, and Zongwei Zhou. "Exploiting Structural Consistency of Chest Anatomy for Unsupervised Anomaly Detection in Radiography Images." TPAMI (2024).

 https://www.cs.jhu.edu/~alanlab/Pubs24/xiang2024exploiting.pdf

报告嘉宾：骆路阳 (香港科技大学)

报告时间：2024年5月22日 (星期三)晚上20:30 (北京时间)

报告题目：Understanding and Learning from Imperfect Medical Data

报告人简介：

Dr. Luyang Luo is a Post-doctoral Fellow at the Department of Computer Science and Engineering, The Hong Kong University of Science and Technology (HKUST). He obtained his Ph.D. and B.E. in the Department of Computer Science and Engineering from The Chinese University of Hong Kong. He has published dozens of papers in top-tiered conferences and journals, including MICCAI, IEEE TMI, MIA, Lancet Digital Health, Nature Communications, Radiology Artificial Intelligence, JMRI, etc. He served as a reviewer for conferences and journals including MICCAI, AAAI, ISBI, IEEE TMI, MIA, TIP, IEEE JBHI, etc. He has also served as a program chair for ICLR 2023 workshop on TML4H, a meta reviewer for ICCV CVAMD 2023 and 4th International Workshop on MMMI, the co-editor for the proceedings of TML4H 2023, and the associate editor for IEEE JBHI special issue on Trustworthy Machine Learning for Health Informatics.

个人主页：

https://llyxc.github.io

报告摘要：

Medical data are often small and contain bias, and the obtained labels are often scarce and diverse. In other words, medical data are imperfect, especially for developing data-hungry AI in the era of deep learning and foundation models. This talk will introduce a series of our works on the understanding of imperfect medical data and how to learn from such data, including omni-supervised learning frameworks for label-efficient model development, trustworthy AI for debiased and explainable diagnosis, and some of our recent exploration on applying AIs for clinical research.

参考文献：

[1] Chai, Z., Luo, L., Lin, H., Heng, P. A., & Chen, H. (2024). Deep Omni-supervised Learning for Rib Fracture Detection from Chest Radiology Images. IEEE TMI, 2024.

[2] Luo, L., Chen, H., Zhou, Y., Lin, H., & Heng, P. A. (2021). OXnet: Deep Omni-Supervised Thoracic Disease Detection from Chest X-rays. MICCAI, 2021.

[3] Luo, L., Huang, X., Wang, M., Wan, Z., & Chen, H. (2024). Medical Image Debiasing by Learning Adaptive Agreement from a Biased Council. arXiv preprint arXiv:2401.11713.

[4] Luo, L., Chen, H., Xiao, Y., Zhou, Y., Wang, X., Vardhanabhuti, V., ... & Heng, P. A. (2022). Rethinking Annotation Granularity for Overcoming Shortcuts in Deep Learning–based Radiograph Diagnosis: A Multicenter Study. Radiology: Artificial Intelligence, 2022.

[5] Luo, L., Xu, D., Chen, H., Wong, T. T., & Heng, P. A. (2022, September). Pseudo Bias-balanced Learning for Debiased Chest X-ray Classification. MICCAI, 2022.

[6] Bie, Y., Luo, L., & Chen, H. (2024). MICA: Towards Explainable Skin Lesion Diagnosis via Multi-Level Image-Concept Alignment. AAAI, 2024.

[7] Luo, L., Wang, X., Lin, Y., Ma, X., Tan, A., Chan, R., ... & Chen, H. (2024). Deep Learning in Breast Cancer Imaging: A Decade of Progress and Future Directions. IEEE Reviews in Biomedical Engineering, 2024.

[8] Xiang, H., Xiao, Y., Li, F., Li, C., Liu, L., Deng, T., ... & Chen, H. (2024). Development and Validation of an Interpretable Model Integrating Multimodal Information for Improving Ovarian Cancer Diagnosis. Nature Communications, 2024.

Panel 嘉宾：韩楚 (广东省人民医院)

嘉宾简介：

韩楚，广东省杰青，博士研究生导师，广东省人民医院特聘副研究员，广东省医学影像智能分析与应用重点实验室PI，香港中文大学计算机科学与工程博士。长期聚焦肿瘤医学图像人工智能算法和肿瘤计算病理等领域的研究。近五年来，在IEEE-TPAMI､IEEE-TNNLS､IEEE-TMI､Medical Image Analysis (MedIA)､CVPR､MICCAI､ECCV等高水平期刊和会议发表学术论文50余篇，其中以第一或通讯作者身份（含共同）发表论文20余篇，申请发明专利20余项（授权6项）。先后获国自然面上基金，国自然青年基金、广东省自然科学基金-杰出青年基金、广东省海外博士后人才支持计划等项目资助，曾获Pacific Graphics最佳学生论文奖，数字病理转化项目获国家卫健委举办的数字健康应用大赛特等奖。目前担任IEEE-TMI､MedIA､IEEE-JBHI等多个高水平期刊和会议审稿人。

个人主页：

https://chuhan89.com/

Panel 嘉宾：张建鹏 (阿里巴巴达摩院)

嘉宾简介：

张建鹏，阿里巴巴达摩院算法专家，浙江大学博士后，研究方向为医学影像智能计算。近五年在IEEE-TPAMI､IEEE-TMI､CVPR､ICCV､MICCAI等本领域顶级期刊/会议发表论文30余篇，其中第一/共同第一/通讯作者论文12篇，谷歌学术引用4000余次，第一作者单篇最高引用800余次，三篇入选ESI高被引。2022～2023连续两年入选美国斯坦福大学John P. A. Ioannidis教授团队发布的年度全球前2%科学家榜单。担任MICCAI 2024 Area Chair，以及IEEE TPAMI､TMI､CVPR､ICCV､MICCAI等二十余个国际期刊和会议的评审工作，多次被评为IEEE TMI､ICCV杰出评审人。

个人主页：

https://jianpengz.github.io/

Panel 嘉宾：夏勇 (西北工业大学)

嘉宾简介：

西北工业大学计算机学院长聘教授、博导、空天地海一体化大数据应用技术国家工程实验室成员。研究方向为医学影像智能计算，近5年在JAMA Network OpenRadiology､IEEE-TPAMI､TMI､TIP､TNNLS､MedIA､NeurIPS､CVPR､ECCV､MICCAI､AAAI､IJCAI发表论文70余篇，谷歌引用1.2万余次，先后在BraTS 2020､KiTS 2021､KiPA 2022､SegRap 2023等10余项国际学科竞赛中获得前三名；担任中国体视学学会理事、中国计算机学院数字医学分会常委、中国图象图形学学会视觉大数据专委会常委、陕西省计算机学会人工智能专委会主任，曾担任IBSI 2017、MICCAI 2019/ 2020、ICASSP 2023等学术会议地区主席或分会主席。 

个人主页：

https://teacher.nwpu.edu.cn/yongxia.html

主持人：陈浩 (香港科技大学)

主持人简介：

陈浩，香港科技大学计算机科学与工程系和化学与生物工程系助理教授，研究兴趣包括可信人工智能，医学图像分析，可解释深度学习等。他领导的人工智能医疗实验室（Smart Lab），专注于可信人工智能技术在医疗领域的前沿研究与转化应用。陈教授于2017年获得香港中文大学博士学位。在MICCAI､IEEE-TMI､MIA､CVPR､ICCV､AAAI､IJCAI､Radiology､Lancet Digital Health､Nature Machine Intelligence､Nature  Communications等顶级期刊和会议发表论文100余篇（谷歌学术引用24500余次，h-index 63），连续入选斯坦福大学全球排名前2%科学家名单。此外，陈教授还具有丰富的工业研究和产业转化经验，拥有二十余项人工智能和图像分析方面专利。曾获得2023年亚洲青年科学家、国家教育部高等学校科学研究优秀成果二等奖、北京市科技进步一等奖、2019年人工智能医学影像顶级会议MICCAI青年科学家影响力奖､Elsevier-MICCAI 最佳论文奖、医学影像与增强现实会议最佳论文奖、福布斯中国30岁以下30位精英等奖项，担任包括IEEE TNNLS､J-BHI､CMIG和Medical Physics等期刊编委，担任2024 ACM MM､MICCAI 2021-2023､MIDL 2022-2024､CVPR 2024等多个人工智能与医学影像分析国际会议的领域主席和程序委员，曾带领团队获得15余项国际医学图像分析的挑战赛冠军。

个人主页：

https://cse.hkust.edu.hk/~jhc/

特别鸣谢本次Webinar主要组织者：

主办AC：陈浩 (香港科技大学)

协办AC：夏勇 (西北工业大学)