More IMPATIENT: A gridding-accelerated Toeplitz-based strategy for non-Cartesian high-resolution 3D MRI on GPUs

• Published in: Journal of parallel and distributed computing Vol. 73; no. 5; pp. 686 - 697
• Main Authors: JIADING GAI, OBEID, Nady, HOLTROP, Joseph L, WU, Xiao-Long, FAN LAM, MAOJING FU, HALDAR, Justin P, HWU, Wen-Mei W, LIANG, Zhi-Pei, SUTTON, Bradley P
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier 01-05-2013
• Subjects: Applied sciences; Artificial intelligence; Computer science; control theory; systems; Computer systems and distributed systems. User interface; Exact sciences and technology; Language processing and microprogramming; Pattern recognition. Digital image processing. Computational geometry; Software; Capability index; Bottleneck; High resolution; MRI; Toeplitz; Measurement sensor; Iterative method; Toeplitz matrix; Nuclear magnetic resonance imaging; GPU; Image reconstruction; CUDA; Non-Cartesian; Software tool; Gridding; Graphic processing unit; Massive parallelism; Data structure; Accelerator; gridding; non-Cartesian
• ISSN: 0743-7315; 1096-0848
• DOI: 10.1016/j.jpdc.2013.01.001

Several recent methods have been proposed to obtain significant speed-ups in MRI image reconstruction by leveraging the computational power of GPUs. Previously, we implemented a GPU-based image reconstruction technique called the Illinois Massively Parallel Acquisition Toolkit for Image reconstruction with ENhanced Throughput in MRI (IMPATIENT MRI) for reconstructing data collected along arbitrary 3D trajectories. In this paper, we improve IMPATIENT by removing computational bottlenecks by using a gridding approach to accelerate the computation of various data structures needed by the previous routine. Further, we enhance the routine with capabilities for off-resonance correction and multi-sensor parallel imaging reconstruction. Through implementation of optimized gridding into our iterative reconstruction scheme, speed-ups of more than a factor of 200 are provided in the improved GPU implementation compared to the previous accelerated GPU code.