Enabling task-based medical image quality assessment via machine learning

Date: 2021/12/09 - 2021/12/09

Academic Seminar: Enabling task-based medical image quality assessment via machine learning

Speaker: Dr. Weimin Zhou, postdoctoral scholar in the Department of Psychological & Brain Sciences at University of California, Santa Barbara (UCSB)

Time: 9:00 a.m.-10:00 a.m., Dec 9th, 2021 (Beijing Time)

Location: via Feishu

Abstract

Biomedical imaging systems should be assessed and optimized via task-based image quality measures. Task-based image quality measures quantify the performance of some observers for specific diagnostic tasks. Model observers, such as the Bayesian Ideal observer and the Hotelling observer, have been advocated to evaluate task-based image quality measures. Machine learning methods have been extensively explored and successfully applied to establish inference models for performing various medical imaging tasks, such as abnormality detection. This talk will introduce several of my developed machine learning methods for enabling task-based image quality assessment in medical imaging. In the first part of my talk, I will describe supervised learning-based methods to approximate the Bayesian Ideal observer and the Hotelling observer. I will focus on binary signal detection tasks and joint signal detection and localization tasks. In the second part of my talk, I will introduce an augmented generative adversarial network (GAN) architecture named advancedAmbientGAN for learning statistical properties of to-be-imaged objectsfrom raw medical imaging measurements. This method enables the establishment of stochastic object models that capture realistic textures and anatomical variations for use in assessing task-based image quality. In the third part of my talk, I will present a GAN-enabled Markov-Chain Monte Carlo (MCMC-GAN) method. This method significantly extends the domain of applicability of MCMC methods for computing the Bayesian Ideal observer.

Biography

Dr. Weimin Zhou is a postdoctoral scholar in the Department of Psychological & Brain Sciences at University of California, Santa Barbara (UCSB). Before joining UCSB in 2020, he was a research assistant in the Department of Biomedical Engineering at Washington University in St. Louis (WashU) and a visiting scholar in the Department of Bioengineering at University of Illinois at Urbana-Champaign (UIUC). Dr. Zhou received his dual BSc(Eng) degrees in Telecommunications Engineering with Management from Beijing University of Posts and Telecommunications (BUPT) and Queen Mary University of London (QMUL) in 2014. He earned his M.S. and Ph.D. degrees in Electrical Engineering at WashU in 2016 and 2020, respectively. Dr. Zhou is the recipient of the SPIE Community Champion and the SPIE Medical Imaging Cum Laude Poster Award (best poster award). His research utilizes various tools to advance biomedical imaging technologies and bridges the fields of biomedical imaging, machine learning, vision science, and image science. His work on machine learning-enabled task-based image quality assessment was published in the top medical imaging journal IEEE TMI.