Abstract:

Improving the speed of MRI has always been a problem to be solved in the field of magnetic resonance. Using multi-coil scan is a common method. However, when the acceleration factor is greater than 4, the image quality obtained by traditional compressed sensing magnetic resonance imaging reconstruction algorithms is not high. In this study, we propose a multi-coil MRI image reconstruction method named as ISTAVS algorithm based on ISTAVS-Net. It combines the ISTA algorithm with the splitting idea of VS-Net network. It is expanded into an ISTAVS-Net network. Each iteration step is combined with the network module，which has higher interpretability than black box U-Net network. The residual mechanism is introduced into the ISTAVS-Net to increase the non-linear expression ability and accuracy. Sparse transformation, shrinkage threshold and regularization parameter are automatically learned during training, which increases the flexibility of reconstruction. The test results of the Globus knee dataset show that the ISTAVS-Net network is better than traditional L1-ESPIRiT and ISTA algorithm. And the image quality and performance metrics are improved than U-Net and ISTA-Net+ and VS-Net at different acceleration factors. It has better ability to recover the tissue details at high acceleration factors. It has good robustness and is more suitable for fast and high-quality reconstruction of data acquired on clinical MR scanners. It can also expand the application range of MRI.

Key words: magnetic resonance imaging, physical model, image reconstruction, multi coil under-sampling, VS-Net, ISTA algorithm