Colloquium: Dr. Il Yong Chun, Ph.D., University of Michigan, Breaking Image Limits

All dates for this event occur in the past.

Dreese 260

 

Abstract

 

“Extreme” computational imaging collects extremely undersampled or inaccurate measurements, and provides significant benefits to diverse imaging applications. The examples in medical imaging include highly undersampled MRI to reduce imaging time, sparse-view CT to reduce radiation dose and cancer risk from CT scanning, etc. However, obtaining an accurate image within a reasonable computing time is challenging in extreme computational imaging.  ~ Convolutional neural network (CNN) is inherently ill-matched to iterative signal recovery con-structed by explicitly formulating objective function. The existing analysis operator learning methods require huge memory to store many overlapping patches, and are not suitable for learning kernels with big data that consists of many large dimensional signals. The first half of this talk introduces my recent 1) convolutional analysis operator learning (CAOL) framework that resolves the two aforementioned problems, 2) new block proximal gradient method using a majorizer for fast and stable CAOL; and shows 3) a benefit of CAOL to sparse-view CT collecting only 12.5%projection measurements.   ~ Nonetheless, the CAOL-based iterative algorithm needs several hundreds of iterations to converge, detracting from its practical use. The second half of this talk introduces my recent 1) recurrent CNN architecture, BCD-Net that unfolds block coordinate descent (BCD) algorithm using CAOL, designed to resolve the convergence issue; and shows 2) a benefit of deep layered BCD-Net to highly undersampled MRI collecting only 10% k-space data.  ~ Making extreme computational imaging practically feasible breaks new ground in providing safe and comfortable medical imaging to patients, and developing reliable early-stage brain abnormality detector.

Bio

Dr. Il Yong Chun received the B.Eng.E.E. degree from Korea University, Seoul, South Korea, and the Ph.D. degree in electrical and computer engineering from Purdue University, West Lafayette, IN, USA, in 2009 and 2015, respectively. From 2015 to 2016, he was a Postdoctoral Research Asso-ciate in Mathematics, Purdue University, West Lafayette, IN, USA. He is currently a Postdoctoral Research Fellow in Electrical Engineering and Computer Science, the University of Michigan, Ann Arbor, MI, USA. His research interests include convolutional operator learning with big data, deep neural networks, compressed sensing, and nonconvex optimization, applied to “extreme” computational imaging and translational neuroimaging.

Category: Seminar