Congratulations to Yankun Xu on his successful PhD dissertation defense

Fig. 1: Congratulations to Yankun Xu on his successful PhD dissertation defense, and thanks for Chair Professor Mohamad Sawan on his supervision of this thesis project！

On the Morning of June 7, a School of Engineering dissertation defense was held in room E2-216, betway必威棋牌登录 Yankun Xu, a PhD candidate of CenBRAIN Neurotech Center of Excellence,successfully defended his doctoral dissertation having the title "Accurate AI-based Early Recognition of Epileptic Seizures from EEG and Video Recordings".

Professor Fei Wufrom Zhejiang University was invited to be the chair of the defense committee,Professor Lin YangandProfessor Xin Yuanfrom Westlake University,Professor Jiuwen CaoandProfessor Jianhai Zhangfrom Hangzhou Dianzi University were the members of the defense committee.Dr. Yun-Hsuan Chen, Research Assistant Professor in Westlake University, was the secretary of this evaluation committee. Several CenBRAIN Neurotech Center of Excellence members attended this defense.

Ph.D Graduate

"I wish to express my deepest gratitude to my advisor, Prof. Mohamad Sawan. His unwavering support, insightful guidance, and constructive criticism have been invaluable. Prof. Sawan's dedication to my academic and personal growth, combined with his exceptional mentoring, has inspired and challenged me to reach new heights." Yankun said.

Dr. Yankun Xu focuses on epileptic seizure detection and prediction.He has introduced several algorithms for improving seizure detection and prediction. Also, he has established collaboration with epilepsy center of SAHZU to focus on video-based seizure detection project. During his PhD study at CenBRAIN Neurotech, he has published more than ten papers as first or co-author. Among these, he has published four papers as the first author on top journals and conferences. Besides, Dr. Yankun received an outstanding PhD student award, and will graduate a PhD degree from Zhejiang University with outstanding mention.

Ph.D Thesis Abstract

As for patients with intractable epilepsy, an early seizure recognition system can provide feasibility for a timely intervention prior to a seizure onset. This thesis focuses on the application scenarios of EEG and video, aiming to address challenges such as insufficient accuracy, long detection latency, imbalanced training data, and lack of video solutions in previous research. With the research objective of "accurate and timely artificial intelligence-based early seizure recognition", the thesis investigates the theory, key technologies, typical architectures, and core algorithms in the field of algorithm-based early seizure recognition. Firstly, this thesis introduces a crossing period between ictal period and interictal period for EEG-based early seizure detection, it proposes a regression model based on probabilistic prediction instead of traditional binary classification model. Secondly, this thesis presents a preictal signal synthesis method based on generative adversarial networks, aiming to address the issue of imbalanced training samples for seizure prediction study. Thirdly, this thesis proposes a comprehensive computer vision-based solution for video-based early seizure detection.