VALSE Webinar 20220706-17期 总第284期 联邦学习在医学图像处理的应用 ...

报告嘉宾：李霄霄 (The University of British Columbia)

报告题目：Federated learning for healthcare: from theory to practice

报告嘉宾：徐子乐 (Nvidia)

报告题目：Techniques and Tools for Collaborative Development of AI Models across Institutes

报告嘉宾：李响 (哈佛大学、麻省总医院)

报告题目：联邦学习在医学图像处理的应用

Panel嘉宾：

李霄霄 (The University of British Columbia)、徐子乐 (Nvidia)、李响 (哈佛大学、麻省总医院)、陈浩 (香港科技大学)、周郁音 (UC Santa Cruz)、秦璟 (香港理工大学)

Panel议题：

1. 像联邦学习这样的学习范式部署是否涉及到当前隐私法规对此类技术解释？

2. 基线联邦学习无法缓解的私人信息泄露威胁有哪些，哪些技术有望解决这些威胁？

3. 您如何看待联邦学习在临床中的优势，还有哪些障碍需要我们克服？

4. 联邦学习还有有哪些尚未被广泛探索或开发的优势？

5. 联邦学习系统如何体现公平和包容性？

*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：李霄霄 (The University of British Columbia)

报告时间：2022年07月06日 (星期三)晚上20:00 (北京时间)

报告题目：Federated learning for healthcare: from theory to practice

报告人简介：

Dr. Xiaoxiao Li is an Assistant Professor at the Department of Electrical and Computer Engineering (ECE) at the University of British Columbia (UBC) starting August 2021. Before joining UBC, Dr. Li was a Postdoc Research Fellow in the Computer Science Department at Princeton University. Dr. Li obtained her PhD degree from Yale University in 2020. Dr. Li’s research interests range across the interdisciplinary fields of deep learning and biomedical data analysis, aiming to improve the trustworthiness of AI systems for healthcare. Dr. Li has had over 30 papers published in leading machine learning conferences and journals, including NeurIPS, ICML, ICLR, MICCAI, IPMI, BMVC, IEEE Transactions on Medical Imaging, and Medical Image Analysis. Her work has been recognized with several best paper awards at international conferences.

个人主页：

https://xxlya.github.io/xiaoxiao/

报告摘要：

With the development of artificial intelligence (AI) for healthcare, people started to pay more attention to the challenges and risks associated with AI models to be used in sensitive healthcare applications. Currently, there are still many practical barriers to the promotion of AI in healthcare applications, such as insufficient training samples, difficulties in data sharing and labeling, etc. To overcome these barriers and accelerate the application of AI in healthcare, we must work on developing more accurate models with more diverse data. AI has emerged as an important trend in AI research and industry.  This presentation will discuss how to build a new generation of AI-powered healthcare systems and focus on the ongoing progress in both theories and practice towards advancing federated learning and other collaborative learning methods in medical data analysis.

参考文献：

[1] Li, Xiaoxiao, et al. "Fedbn: Federated learning on non-iid features via local batch normalization." arXiv preprint arXiv:2102.07623 (ICLR 2021).

[2] Huang, Baihe, et al. "Fl-ntk: A neural tangent kernel-based framework for federated learning analysis." International Conference on Machine Learning. PMLR, 2021.

[3] Peng, Liang, et al. "Fedni: Federated graph learning with network inpainting for population-based disease prediction." arXiv preprint arXiv:2112.10166 (TMI 2021).

[4] Li, Xiaoxiao, et al. "Multi-site fMRI analysis using privacy-preserving federated learning and domain adaptation: ABIDE results." Medical Image Analysis 65 (2020): 101765.

[5] Lu, Nan, et al. "Federated Learning from Only Unlabeled Data with Class-Conditional-Sharing Clients." arXiv preprint arXiv:2204.03304 (ICLR 2022).

报告嘉宾：徐子乐 (Nvidia)

报告时间：2022年07月06日 (星期三)晚上20:30 (北京时间)

报告题目：Techniques and Tools for Collaborative Development of AI Models across Institutes

报告人简介：

Dr. Ziyue Xu is currently a Senior Scientist at Nvidia Corp., his research interests lie in the area of image analysis and machine learning with applications in biomedical and clinical imaging. Before joining Nvidia, he was a Staff Scientist in the Center for Infectious Disease Imaging in National Institutes of Health. Dr. Xu obtained B.S. in Electronics Engineering from Tsinghua University, and M.S./ Ph.D. in Computer Engineering from the University of Iowa. He is an Associate Editor for the journals, TMI, JBHI, CBM, and CMIG. He also serves as a regular reviewer for a list of top-ranking journals, and program chair/ committee member for various conferences in this field.

个人主页：

https://sites.google.com/site/xuziyue/home

报告摘要：

Federated learning (FL)is a method used for training artificial intelligence models with data from multiple sources while maintaining data anonymity, thus removing many barriers to data sharing. In this talk, we will discuss two major aspects: the research towards personalized FL models, and the tools needed to perform a real life multi-institute FL study. Specifically, the Federated Super Model algorithm will be discussed, together with the EXAM (electronic medical record (EMR)chest X-ray AI model)study, which predicts the future oxygen requirements of symptomatic patients with COVID-19 using inputs of vital signs, laboratory data and chest X-rays from 20 institutes across the globe with the help of NVFlare - the NVIDIA Federated Learning Application Runtime Environment.

参考文献：

[1] Dayan, Ittai, et al. "Federated learning for predicting clinical outcomes in patients with COVID-19." Nature medicine 27.10 (2021): 1735-1743.

[2] Xu, An, et al. "Closing the Generalization Gap of Cross-silo Federated Medical Image Segmentation." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022.

报告嘉宾：李响 (哈佛大学、麻省总医院)

报告时间：2022年07月06日 (星期三)晚上21:00 (北京时间)

报告题目：联邦学习在医学图像处理的应用

报告人简介：

李响博士毕业于上海交通大学自动化系， 获美国佐治亚大学计算机科学博士 (生物医疗影像方向)，师从佐治亚大学杰出教授，美国医学与生物工程院 (AIMBE)会士刘天明教授。2016年毕业后任哈佛大学医学院和麻省总医院博士后，接受美国医学院院士，前麻省总医院放射科主任James Thrall和麻省总医院先进医学计算和分析中心主任李全政教授的指导。主要从事机器学习，计算机视觉及大数据科学在医学影像中的应用研究和算法开发。李响博士在医学影像，数据科学，人工智能和机器学习方面期刊及国际顶级会议上发表了百余篇学术论文。担任包括NeurIPS，ICML，ICLR，Nature Communication，Pattern Recognition在内的学术会议和期刊审稿人，及包括Frontiers in Computational Neuroscience，BMC Bioinformatics，Frontiers in Oncology在内的学术期刊编辑，并于2019年起创办了International Workshop on Multiscale Multimodal Medical Imaging会议。其在多模态医学影像融合，以及基于强化学习的临床决策支持系统的多项研究获得来自美国国立卫生研究院 (NIH)的资助，并获得包括MGH Thrall Innovation Grants Award，NVIDIA Global Impact Award，ISBI Best Paper Award在内的多项奖项。

个人主页：

https://xiangli-shaun.github.io/

报告摘要：

联邦学习 (FL)是一种使用来自多个来源的数据训练人工智能模型，同时保持数据隐私的框架，从而可以消除在数据共享，特别是跨区域医学数据共享中的诸多障碍。在这项由麻省总医院，Nvidia以及全世界20余家医院和医学机构的合作研究中，我们使用多中心数据来训练一个称为 EXAM (电子病历EMR+胸部X射线影像+AI模型)的联邦学习模型。EXAM模型使用包括体征，血液检查，和胸部 X射线影像在内的多模态合并特征作为输入，对COVID-19确诊和疑似患者在24/72小时内的健康状况和辅助呼吸装置需求进行预测。通过联邦学习机制，各个参与的机构可以在不分享自身数据的前提下更有效地进行模型协同训练：与独立进行模型训练相比，EXAM使得各个机构的平均 AUC 提高了16%，通用性提高了38%。EXAM 对预测患者24/72小时内健康风险的平均AUC达到了0.92以上。在进一步进行的独立中心验证中，EXAM对于重症 (24小时内需要机械呼吸机或死亡)的预测敏感性达到了0.95，特异性达到0.88，说明了其在实际使用中的临床价值。在这次国际大规模联邦学习研究中，我们在没有数据交换的情况下促进了各个中心的人工智能平台开发协作，开发的EXAM模型在异构、未归一化的数据中进行了有效的泛化，为联邦学习在医疗数据科学中的更广泛使用奠定了基础。

参考文献：

[1] Dayan, Ittai, et al. "Federated learning for predicting clinical outcomes in patients with COVID-19." Nature medicine 27.10 (2021): 1735-1743.

[2] Zhong, Aoxiao, et al. "Deep metric learning-based image retrieval system for chest radiograph and its clinical applications in COVID-19." Medical Image Analysis 70 (2021): 101993.

Panel嘉宾：陈浩 (香港科技大学)

嘉宾简介：

陈浩博士是香港科技大学计算机科学与工程系助理教授。他领导的人工智能医疗实验室 (Smart Lab)，专注于可信赖人工智能技术在医疗领域的研究与应用。2017年获得香港中文大学博士学位，发表学术论文100余篇 (谷歌学术引用次数14300余次，h-index 53)，包括MICCAI、IEEE-TMI、MIA、CVPR、AAAI、Lancet Digital Health、Nature Machine Intelligence、JAMA等。此外，陈博士还具有丰富的工业研究经验，拥有十余项人工智能和图像分析专利。陈博士曾获得2019年医学影像顶级会议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Frontiers in Artificial Intelligence、Frontiers in Big Data和Medical Physics等期刊副主编，担任MICCAI 2021-2022、ISBI 2022、MIDL 2022、AAAI 2022 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/

Panel嘉宾：周郁音 (UC Santa Cruz)

嘉宾简介：

Dr. Yuyin Zhou is an Assistant Professor of Computer Science and Engineering at UC Santa Cruz. She received her Ph.D. from the Computer Science Department at Johns Hopkins University in 2020 and was a postdoctoral researcher at Stanford University from 2020 to 2021. Yuyin’s research interests span the fields of medical image computing, computer vision, and machine learning, especially the intersection of them. She has published many papers at top-tier conferences and journals including CVPR, ICCV, AAAI, TPAMI, TMI, MedIA, etc. Yuyin Zhou has led the ICML 2021 workshop on Interpretable Machine Learning in Healthcare, the ICCV 2021 workshop on Computer Vision for Automated Medical Diagnosis, and co-organized ML4H 2021, the 9th CVPR MCV workshop. She served as a senior program committee for IJCAI 2021 and AAAI 2022, an area chair for MICCAI 2022, CHIL 2022.

个人主页：

https://yuyinzhou.github.io/

Panel嘉宾：秦璟 (香港理工大学)

嘉宾简介：

QIN, Jing (Harry)is currently an associate professor in Centre for Smart Health, School of Nursing, The Hong Kong Polytechnic University. His research focuses on creatively leveraging advanced virtual reality (VR)and artificial intelligence (AI)techniques in healthcare and medicine applications and his achievements in relevant areas has been well recognized by the academic community. He won the Hong Kong Medical and Health Device Industries Association Student Research Award for his PhD study on VR-based simulation systems for surgical training and planning. He won 5 best paper awards for his research on AI-driven medical image analysis and computer-assisted surgery. He served as a local organization chair for MICCAI 2019, technical program committee (TPC)members for many academic conferences, speakers for many invited talks, and referees for many prestigious journals in relevant fields.

个人主页：

https://research.polyu.edu.hk/en/persons/jing-qin

主持人：雷柏英 (深圳大学)

主持人简介：

雷柏英，国家高层次人才入选者，深圳大学特聘教授， (留学)博士生导师，深圳市海外高层次人才 (孔雀计划)、深圳市高层次后备级人才，深圳市孔雀团队核心成员等，获新加坡南洋理工大学博士学位。先后在美国北卡大学教堂山分校和法国计算和自动化研究所等研究机构进行研究和访问。主要研究方向为医学图像处理和人工智能。在IEEE TMI、IEEE TNNLS、Medical Image Analysis 以第一/通讯作者 (含共同)发表SCI论文102篇 (19篇IEEE汇刊、1篇ESI高被引)。谷歌学术总引用4666次，单篇文章最高引用321次，H指数32。获授权专利22项 (3项已转让)。主持国家自然科学基金面上项目等共19项。现任IEEE TNNLS、IEEE TMI、Medical Image Analysis、Neural Computing & Application 编委。现为IEEE 高级会员，IEEE Bio Imaging Signal Processing (BISP)Technical Committee (TC)技术委员会委员 (中国1人)，Biomedical Imaging and Image Processing (BIIP)TC技术委员会委员 (中国2人)，医学图像顶级学术会议MICCAI 2021、MICCAI 2022领域主席，IEEE Guangzhou Section, Women in Engineering Affinity Group 主席，人工智能A类会议AAAI (2019，2020)、IJCAI (2019，2020)程序委员会委员、中国人工智能学会 (CAAI)青工委委员、模式识别与人工智能专委、计算机视觉专委、女科技工作者工作委员，阿尔茨海默病防治协会人工智能技术专委、中国图像图形学学会 (CSIG)人工智能专委、脑图谱专委、女科技工作者工作委员、青工委委员，生物医学工程学会青工委委员。指导学生获MICCAI国际竞赛3项任务冠军。获吴文俊人工智能科学技术奖三等奖 (排名第3)，深圳市科学技术奖一等奖 (排名第3)。2020年和2021年入选美国斯坦福大学发布的“全球前2%顶尖科学家”。2021年入选全球顶尖前10万科学家。

个人主页：

http://bme.szu.edu.cn/20181/0612/66.html

特别鸣谢本次Webinar主要组织者：

主办AC：雷柏英 (深圳大学)

协办AC：陈浩 (香港科技大学)