Computational Magnetic Resonance Imaging: Combining Math, Physics, and Computation to Improve Healthcare

Monday, January 28, 2019
11:00 AM to 12:00 PM
EER 3.646
Free and open to the public

Medical imaging modalities such as MRI have transformed modern diagnostic medicine by providing safe and noninvasive approaches to seeing inside the body. However, traditional designs treat the data acquisition separately from the data processing, thereby limiting their capabilities. For example, conventional MRI exams do not leverage shared information across different scans. In this talk I present computational approaches to medical imaging that consider joint design of the data collection and the reconstruction algorithm to overcome these limitations. In particular, I will show how these ideas can be used to streamline the MRI exam and reduce its cost without compromising diagnostic image quality. By incorporating imaging physics and phenomenological constraints into the reconstruction, we are able to model and recover the rich signal and physiological dynamics captured in the acquisition. The reconstruction algorithm in turn informs a redesigned sampling scheme to accelerate the acquisition. Using these techniques, I show clinical integration of fast pediatric MRI exams that has led to over 3000 patient scans, a two-fold reduction in exam times, and a three-fold reduction in exam costs. I also present efficient numerical implementations to maintain clinically acceptable latencies. Finally, I discuss approaches to leveraging the previous accelerated clinical scans to learn new acquisition and reconstructions schemes that are tailored to the imaging task. The computational spin to MRI provides a blueprint for applying similar concepts to other biomedical and engineering systems.


Jonathan Tamir

Jonathan Tamir

University of California, Berkeley

Jon Tamir is a research associate at the University of California, Berkeley, where he also recently completed his PhD in electrical engineering and computer sciences. Before coming to Berkeley, he received the BS degree in electrical engineering from The University of Texas at Austin. His research interests include computational magnetic resonance imaging, machine learning for inverse problems, and clinical translation. His PhD work focused on applying fast imaging and efficient reconstruction techniques to MRI, with the goal of enabling real clinical adoption. He is a core developer of the Berkeley Advanced Reconstruction Toolbox, available at