Compressed Sensing MRI with Multi-Channel Data Using Multi-Core Processors

Compressed sensing (CS) has emerged as a promising method in the field of magnetic resonance imaging. Taking advantage of the signal sparsity in certain domain via L-1 minimization, CS requires only reduced k-space data to reconstruct an image. Since most clinical MRI scanners are equipped with multi-channel receiver systems, integrating CS with multi-channel systems may not only shorten the scan time but provide a better image quality. However, significant computation time is required to perform CS reconstruction. Furthermore, this burden will be scaled by the number of channels. In this paper, we proposed a reconstruction procedure, which uses multi-core processors to accelerate CS reconstruction from multiple channel data. The performance was tested in terms of comparing to different image sizes and using different number cores of CPU. Experimentally, it shows that the maximum efficiency benefits from parallelizing the CS reconstructions, pipelining multi-channel data on multi-core processors and choosing the numbers of channels as multiple numbers of cores.