GRAPPA-GANs for Parallel MRI Reconstruction
This work provides an incremental improvement in MRI reconstruction speed and quality for medical imaging specialists.
This paper addresses the problem of accelerating MR image acquisitions by combining GRAPPA with a conditional generative adversarial network (GAN) for k-space undersampling reconstruction. The proposed GRAPPA-GAN model improved PSNR from 33.88 (regularized GRAPPA) to 37.65 for an acceleration rate of R=4, consistently outperforming GRAPPA across various acceleration rates.
k-space undersampling is a standard technique to accelerate MR image acquisitions. Reconstruction techniques including GeneRalized Autocalibrating Partial Parallel Acquisition(GRAPPA) and its variants are utilized extensively in clinical and research settings. A reconstruction model combining GRAPPA with a conditional generative adversarial network (GAN) was developed and tested on multi-coil human brain images from the fastMRI dataset. For various acceleration rates, GAN and GRAPPA reconstructions were compared in terms of peak signal-to-noise ratio (PSNR) and structural similarity (SSIM). For an acceleration rate of R=4, PSNR improved from 33.88 using regularized GRAPPA to 37.65 using GAN. GAN consistently outperformed GRAPPA for various acceleration rates.