Calibrationless Parallel Dynamic MRI with Joint Temporal Sparsity

• Published in: Medical Computer Vision: Algorithms for Big Data pp. 95 - 102
• Main Authors: Yu, Yang, Yan, Zhennan, Feng, Li, Metaxas, Dimitris, Axel, Leon
• Format: Book Chapter
• Language: English
• Published: Cham Springer International Publishing 2016
• Series: Lecture Notes in Computer Science
• Subjects: Coil Sensitivity; Dynamic Magnetic Resonance Imaging; Magnetic Resonance Imaging Image; Sensitivity Profile; Sparse Constraint
• ISBN: 3319420151; 9783319420158
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-319-42016-5_9

In this paper, we propose a novel calibrationless method for parallel dynamic magnetic resonance imaging (MRI) reconstruction, which overcomes the limitations posed by traditional MRI reconstruction methods that require accurate coil calibration. Thus, calibrationless methods, which remove the requirement of coil sensitivity profiles for MRI reconstruction, are suitable for dynamic MRI. Dynamic MRI contains rich temporal redundant information, i.e., the pixel intensities change smoothly over time. This property can be modeled as various types of temporal sparse priors, in the Fourier transform domain, or in the image domain using finite differences. In addition, the temporally changing patterns of pixels are similar in the various coils, since their signals are different due to the coil sensitivity profiles. Therefore, we model the parallel dynamic MRI problems as joint temporal sparsity tasks, and develop a class of algorithms to solve them efficiently. Experiments on parallel dynamic MRI datasets demonstrate that our proposed methods outperform the state-of-the-art parallel MRI reconstruction algorithms.