MR Image Reconstruction as a Computational Imaging Problem: From Model-Based Reconstruction and Sparsity to Machine Learning

Share

Over the past two decades, image reconstruction has tremendously gained in importance in MRI enabling reduced scan time, improved image quality, and extracting additional information from the measurements. In this time, MRI has witnessed extensive developments in advanced computational algorithms for image reconstruction, many of which have been fueled by signal processing advances in several areas, including multi-channel sampling, compressive sensing, dictionary learning, low-rank, and structured low-rank methods. Recently, also neural networks have been employed for image reconstruction achieving further improvements in scan time and image quality. Most importantly, some of these techniques have found their way in the products of MRI vendors and show significant impact in the clinical practice. These developments, together with the advancements in computational hardware have opened a new research field of MRI reconstruction as a computational imaging problem. In this talk, I will explain the framework of MRI reconstruction as a computational imaging problem and discuss some of the advantages it gives in addressing important clinical needs in MRI.

 



  Date and Time

  Location

  Hosts

  Registration



  • Date: 12 Oct 2022
  • Time: 12:00 PM to 01:00 PM
  • All times are (GMT-05:00) US/Eastern
  • Add_To_Calendar_icon Add Event to Calendar

ZOOM Meeting Link

 https://fdu.zoom.us/j/93815826146

 

Meeting ID: 938 1582 6146

 

  • 1000 River Road
  • Teaneck , New Jersey
  • United States 07666
  • Building: Muscarelle Center, M105,
  • Room Number: M105

  • Co-sponsored by Fairleigh Dickinson University
  • Starts 18 July 2022 12:00 PM
  • Ends 12 October 2022 01:00 PM
  • All times are (GMT-05:00) US/Eastern
  • No Admission Charge


  Speakers

Dr. Mariya Doneva of Philips Research, Hamburg, Germany

Topic:

MR Image Reconstruction as a Computational Imaging Problem: From Model-Based Reconstruction and Sparsity to Machine Lear

Over the past two decades, image reconstruction has tremendously gained in importance in MRI enabling reduced scan time, improved image quality, and extracting additional information from the measurements. In this time, MRI has witnessed extensive developments in advanced computational algorithms for image reconstruction, many of which have been fueled by signal processing advances in several areas, including multi-channel sampling, compressive sensing, dictionary learning, low-rank, and structured low-rank methods. Recently, also neural networks have been employed for image reconstruction achieving further improvements in scan time and image quality. Most importantly, some of these techniques have found their way in the products of MRI vendors and show significant impact in the clinical practice. These developments, together with the advancements in computational hardware have opened a new research field of MRI reconstruction as a computational imaging problem. In this talk, I will explain the framework of MRI reconstruction as a computational imaging problem and discuss some of the advantages it gives in addressing important clinical needs in MRI.

Biography:

Dr. Mariya Doneva is a Senior Scientist and a Team Lead at Philips Research, Hamburg, Germany, which she joined in 2010. She received her BSc and MSc degrees in Physics from the University of Oldenburg in 2006 and 2007, respectively and her PhD degree in Physics from the University of Lübeck in 2010. She was a Research Associate at Electrical Engineering and Computer Sciences Department at UC Berkeley between 2015 and 2016. Her work has yielded many innovations related to imaging workflow improvements, novel quantitative MRI approaches, and most prominently fast MRI data acquisition based on compressed sensing allowing significant reduction of the scan time of routine clinical scans, which has been already integrated in the clinical routine of many hospitals and used to scan millions of patients. She has been granted over 30 patents for her work in MR imaging.

Dr. Doneva was an Organizing Committee Member of multiple conferences including the International Society for Magnetic Resonance in Medicine (ISMRM) (2019-2021), IEEE International Symposium on Biomedical Imaging (ISBI) (2020), the ISMRM Workshop on Data Sampling and Image Reconstruction (2020), and the SIAM Conference on Imaging Science 2022.

She was Guest Editor, IEEE Signal Processing Magazine Special Issue on Computational MRI: Compressive Sensing and Beyond; Editor, comprehensive reference book on Quantitative Magnetic Resonance Imaging; Editorial Board Member, Magnetic Resonance in Medicine and IEEE Transactions on Computational Imaging; and Editor of a reference book on MR image reconstruction.

Dr. Doneva’s research interests include methods for efficient data acquisition, image reconstruction and quantitative parameter mapping in the context of magnetic resonance imaging. Her work involves developing mathematical optimization and signal processing approaches that aim at improving the MR scan efficiency and obtaining robust and reliable (multi-parametric) quantitative information for diagnostics and therapy follow up.

 

Address:United States





Agenda

Over the past two decades, image reconstruction has tremendously gained in importance in MRI enabling reduced scan time, improved image quality, and extracting additional information from the measurements. In this time, MRI has witnessed extensive developments in advanced computational algorithms for image reconstruction, many of which have been fueled by signal processing advances in several areas, including multi-channel sampling, compressive sensing, dictionary learning, low-rank, and structured low-rank methods. Recently, also neural networks have been employed for image reconstruction achieving further improvements in scan time and image quality. Most importantly, some of these techniques have found their way in the products of MRI vendors and show significant impact in the clinical practice. These developments, together with the advancements in computational hardware have opened a new research field of MRI reconstruction as a computational imaging problem. In this talk, I will explain the framework of MRI reconstruction as a computational imaging problem and discuss some of the advantages it gives in addressing important clinical needs in MRI.