Lectures on MR > Courses 2017 > Parellel imaging: Basic and advanced reconstruction concepts
Parellel imaging: Basic and advanced reconstruction concepts

July 20-22, 2017

Course venue:

University Medical Center Göttingen

Course organiser:
Felix Breuer
Magnetic Resonance Bavaria (MRB)

Local organiser:
Martin Uecker
University Medical Center Göttingen/DE

Preliminary faculty:
M. Blaimer, F. Breuer, M. Doneva, J. Hajnal, F. Knoll, S. Kozerke, K. Prüssmann, N. Seiberlich, M. Uecker

Goals of the course:
The course on Parallel imaging: Basic and advanced reconstruction concepts is designed to provide a firm conceptual and practical foundation. Attendees will be brought up to date with established techniques and will develop an appreciation of emerging technologies and methods in multi-channel MRI. The three-day course will rely heavily on interactive tutorials using the MATLAB/Python programming environment. Computers and licenses will be provided for the length of the course. At the end of the course, attendees will understand the basic principles and practical implementation of Cartesian and non-Cartesian parallel imaging methods, spatio-temporal undersampling methods and Compressed Sensing. Attendees will also appreciate the role of these methods in established research practice and how such methods may develop and influence MRI research and practise in the future.

This course will focus on
•Image domain pMRI – reconstruction methods
•k-space pMRI – reconstruction methods
•Artifacts & pitfalls in parallel imaging
•Coils and calibration – practical implementation
•Iterative methods & non-Cartesian reconstruction
•Advanced parallel imaging strategies
•Spatio-temporal undersampling and reconstruction
•Compressed Sensing (CS)
•Future directions in multi-channel MRI

Educational levels:
This course is intended for MR physicists, other scientists and PhD students who already have experience in basic MR methods and knowledge of MR acquisition principles, and who wish to extend their knowledge on Parallel imaging principles and techniques. Some knowledge of MATLAB/Python will be advantageous. All tutorials will be based around pre-existing code prepared for this course. Attendees must have some programming experience and be willing to work with MATLAB/Python.
This course runs from introductory to advanced methods over the three days. At the end of these three days, attendees will take with them the programme code that has been provided and developed by them. This code, in combination with notes taken at the course, will form a package that will enable attendees to implement all the methods discussed during the course.

Course description:
This course is designed to provide a strong practical foundation in the principles of parallel magnetic resonance imaging. Parallel imaging (PI) is now an integral part of many clinical MRI exams. The concepts and methods of PI are informing research in many disparate aspects of MRI. This course is aimed at PhD students and scientists new to parallel imaging who wish to gain a working knowledge of parallel magnetic resonance to underpin their work. The course will be split in two parts, with approximately half the time spent attending lectures and the other half doing practical MATLAB tutorial exercises. We will provide computers and software licenses for the duration of the course.

The course will cover image reconstruction from multiple coils starting with an image domain view (e.g. SENSE) and moving quickly to a k-space perspective (e.g. GRAPPA). We will then look at more advanced methods; non-Cartesian parallel imaging and many of the mathematical tools used in these reconstruction algorithms. We will look at allied methods, in particular spatio-temporal undersampling and subsequent reconstructions. In addition, an introduction to the Compressed Sensing (CS) concept will be given. Finally we will look to the future and discuss how multi channel MRI may impact future directions in MRI.

An integral part of the course will be the MATLAB/Python tutorials where attendees will be able to work through example code provided for them. These examples will demonstrate and enhance their understanding of the concepts discussed throughout the course. Exercises will be set where attendees will modify this code to develop new examples and functionality. At the end of the course they will be free to take this code away with them.

Some previous exposure to MATLAB/Python is preferable but not mandatory. All participants should have some programming experience. All participants will be expected to know essential MR physics. A working knowledge of image acquisition methods and k-space is essential.

Learning objectives:
Image domain parallel imaging
•Define the basic reconstruction problem
•Reconstruct full images from aliased images
•Explore the effects of coil coupling on the reconstruction
•Calculate and measure reconstruction quality

k-space parallel imaging
•Relate image domain and k-space methods
•Assess costs and benefits of image domain and k-space methods
•Calculate and measure reconstruction quality

Coils and calibration
•Understand how coil calibration is achieved
•Compare auto-calibration and pre-calibration approaches
•Establish design criteria for parallel imaging array coils
•Demonstrate how coil calibration errors affect reconstruction

Non-Cartesian parallel imaging
•Define the reconstruction problem
•Review mathematical methods used in reconstruction
•Reconstruct non-uniformly sampled data with iterative methods (CG SENSE)
•Reconstruct non-uniformly sampled data in k-space

Reconstruction with prior knowledge
•Learn about Compressed Sensing
•Understand spatio-temporal undersampling and reconstruction methods
•Calculate and measure reconstruction quality
printversion printversion   |   back to top top

Registration details

supported by: