Abstract:

Modern MRI machines are highly versatile, enabling the in-vivo visualization of various biophysical parameters of the tissue. However, the slow nature of image acquisition introduces several inconvenient tradeoffs, including long scan time, low temporal resolution, and presence of artifacts resulting from patient or physiological motion. These tradeoffs result in several challenges in cardiovascular applications on subjects, who have difficulty holding their breath. In this talk, Dr. Jacob will introduce a learning-based algorithm to overcome these problems. The basic idea is to learn and exploit the significant structure in the data to reduce the data-demand. Dr. Jacob will introduce self-learning strategies, where the structure is learned from the measured data itself, as well as exemplar schemes that learn the structure from training data. The talk will briefly summarize their recent work in cardiac MRI and neuro-imaging, which is available at https://research.engineering.uiowa.edu/cbig/content/publications

Publications | Computational Biomedical Imaging Group | research.engineering.uiowa.edu

S. Poddar, “Joint recovery of high dimensional signals from noisy and undersampled measurements using fusion penalties,” Ph.D Thesis, Department of Electrical and Computer Engineering, University of Iowa, 2018.

I. Bhattacharya, “Pushing the limits of spectroscopic imaging using low-rank based reconstruction,” Ph.D Thesis, Department of Electrical and Computer Engineering, University of Iowa, 2018.