Lecture Review | Local AI Innovation in Mobile and Implantable Medical Devices
Published on: September 29, 2023 | Views: 1424
On September 27, 2023, from 12:30 to 13:30 Dr. Yiyu Shi, a tenured professor in the Department of Computer Science at the University of Notre Dame, Director of the Notre Dame Center for Sustainable Intelligent Computing in Natural Sciences, and visiting scientist at Boston Children’s Hospital, was invited by Sichuan University-Pittsburgh Institute (SCUPI) to deliver a highly anticipated lecture entitled “On-Device AI to Improve Mobile and Implantable Devices in Healthcare” on Jiang ’an campus.
The lecture was moderated by Professor Kang Li, a tenured professor, and Associate Dean for Research of SCUPI. Professor Shijie Bian, Fengquan Lai, Yang Liu, Guangwu Qian, Xiaomei Tan, Jeungphill Hanne and other professors attended the lecture. Before the lecture began, Professor Li provided a brief overview of Dr. Yiyu Shi’s academic career and key research achievements.
Attentive audience during the lecture
To begin with, Dr. Yiyu Shi elaborated extensively on the current state of cardiovascular diagnosis and treatment worldwide. Using charts and data, he highlighted the prevalence rate and severity of cardiovascular diseases globally, emphasizing the necessity, complexity, and challenges of using artificial intelligence to assist in cardiovascular healthcare.
Dr. Yiyu Shi during his presentation
Building upon this foundation, Dr. Yiyu Shi discussed three representative research projects from his work to demonstrate the feasibility of applying artificial intelligence to cardiovascular healthcare. Firstly, he introduced a distribution-based random neural network known as ICA-UNet. Dr. Shi explained in detail how ICA-UNet significantly enhances real-time image segmentation during cardiac interventions by providing real-time segmentation navigation through magnetic resonance imaging, greatly improving computational throughput to meet the demands of real-time segmentation. This approach demonstrated significantly higher accuracy compared to existing algorithms, consequently enhancing the efficiency and success rates of cardiac interventions.
Dr. Yiyu Shi explaining concepts to the audience
Next, Dr. Yiyu Shi presented a novel approach to utilizing 3D modeling and imaging techniques to aid in the diagnosis of congenital heart diseases. He proposed using a 3D neural network for initial cardiac chamber segmentation, followed by layer-specific segmentation using 2D techniques, which were then fused and refined to produce a comprehensive result. Additionally, Dr. Shi presented solutions for identifying spatial structural changes in congenital heart disease images.
Finally, Dr. Yiyu Shi employed regression prediction based on clinical data and deep learning methods using CT imaging to demonstrate the potential assistance of CT images in predicting postoperative outcomes in congenital heart diseases, particularly in cases related to total anomalous pulmonary venous connection (TAPVC). This research provides a scientific basis for modeling the recurrence of pulmonary vein obstruction post-TAPVC surgery.
Towards the end of the lecture, Dr. Yiyu Shi allocated time for a Q&A session. Students enthusiastically asked questions, ranging from personal development inquiries to discussions on cutting-edge research, enhancing their academic knowledge through direct interaction with the professor.
Students actively engaging in Q&A sessions
Dr. Yiyu Shi’s dynamic presentation and insightful content left a profound impression on the faculty and students, offering them a clearer understanding of the latest applications of AI technology in the medical field. The lecture expanded students’ research horizons and ignited their passion for scientific study. With this, the lecture came to a successful conclusion.