Multi-Coil MRI Image Reconstruction Based on ISTAVS-Net of Physical Model

doi:10.11938/cjmr20243109

Abstract

Abstract:

How to improve the speed of MRI is a standing problem in the field of magnetic resonance. A commonly used approach for acceleration is multi-coil scan. However, when the acceleration factor exceeds 4, the image quality obtained by traditional compressed sensing magnetic resonance imaging (CS-MRI) reconstruction algorithms becomes unsatisfactory. In this study, we propose a multi-coil MRI image reconstruction method named as ISTAVS algorithm based on physical model. It combines the ISTA algorithm with the splitting idea of VS-Net, and is expanded into an ISTAVS-Net. Each iteration step is combined with the network module, which has higher interpretability than black box U-Net. 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 outperforms traditional L1-ESPIRiT and ISTA algorithm by multiple indicators, including the improvement of image quality and performance metrics over U-Net, ISTA-Net+ and VS-Net at different acceleration factors, and the ability to recover tissue details at high acceleration factors. The proposed network demonstrates good robustness and is more suitable for fast and high-quality reconstruction of data acquired on clinical MR scanners, thereby can expand the application range of MRI.

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

CLC Number:

HUANG Min, ZHU Junlin, KAO Yuchen, ZHOU Dao, TANG Qiling. Multi-Coil MRI Image Reconstruction Based on ISTAVS-Net of Physical Model[J]. Chinese Journal of Magnetic Resonance, 2024, 41(4): 418-429.

Figures/Tables 8

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Table 1

Fig. 5

Table 2

Fig 6

References 25

[1]	LUSTIG M, DONOHO D, PAULY J M. Sparse MRI: The application of compressed sensing for rapid MR imaging[J]. Magn Reson Med, 2007, 58(6): 1182-1195. doi: 10.1002/mrm.21391 pmid: 17969013
[2]	SONG Y, SHEN L, XING L, et al. Solving inverse problems in medical imaging with score-based generative models[J]. arXiv preprint, arXiv: 2111.08005, 2021.
[3]	GENG C H, JIANG M F, FANG X, et al. HFIST-Net: High-throughput fast iterative shrinkage thresholding network for accelerating MR image reconstruction[J]. Comput Meth Prog Bio, 2023, 232: 107440.
[4]	PAL A, RATHI Y. A review and experimental evaluation of deep learning methods for MRI reconstruction[J]. J Mach Learn Biomed Imaging, 2022, 1: 001.
[5]	WANG S, SU Z, YING L, et al. Accelerating magnetic resonance imaging via deep learning[C]// 2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI), Prague, Czech Republic: IEEE, 2016: 514-517.
[6]	JIN K H, MCCANN M T, FROUSTEY E, et al. Deep convolutional neural network for inverse problems in imaging[J]. IEEE Trans Image Process, 2017, 26(9): 4509-4522.
[7]	ELIZABETH C, JOSEPH C, JOHN P, et al. Analysis of deep complex-valued convolutional neural networks for MRI reconstruction and phase-focused applications[J]. Magn Reson Med, 2021, 86(2):1093-1109. doi: 10.1002/mrm.28733 pmid: 33724507
[8]	ZHAO X, YANG T, LI B, et al. SwinGAN: A dual-domain swin transformer-based generative adversarial network for MRI reconstruction[J]. Comput Biol Med, 2023, 153: 106513.
[9]	LI Y J, YANG X Y, YANG X M. Magnetic resonance image reconstruction of multi-scale residual Unet fused with attention mechanism[J]. Chinese J Magn Reson, 2023, 40(3): 307-319.
	李奕洁, 杨馨雨, 杨晓梅. 融合注意力机制的多尺度残差Unet的磁共振图像重建[J]. 波谱学杂志, 2023, 40(3): 307-319. doi: 10.11938/cjmr20223040
[10]	SHI W C, JIN Z Y, YE Z. Fast multi-channel magnetic resonance imaging based on PCAU-Net[J]. Chinese J Magn Reson, 2023, 40(1): 39-51.
	施伟成, 金朝阳, 叶铮. 基于PCAU-Net的快速多通道磁共振成像方法[J]. 波谱学杂志, 2023, 40(1): 39-51. doi: 10.11938/cjmr20222992
[11]	YANG Y, SUN J, LI H B, et al. Deep ADMM-Net for compressive sensing MRI[C]// Advances in neural information processing systems, Barcelona, Spain, 2016: 10-18.
[12]	HAMMERNIK K, KLATZER T, KOBLER E, et al. Learning a variational network for reconstruction of accelerated MRI data[J]. Magn Reson Med, 2018, 79(6): 3055-3071. doi: 10.1002/mrm.26977 pmid: 29115689
[13]	SRIRAM A, ZBONTAR J, MURRELL T, et al. End-to-end variational networks for accelerated MRI reconstruction[J]. arXiv preprint, arxiv: 2004.06688, 2020.
[14]	CHEN E Z, WANG P, CHEN X, et al. Pyramid convolutional RNN for MRI image reconstruction[J]. IEEE Trans Med Imag, 2022, 41(8): 2033-2047.
[15]	QIAO X, HUANG Y, LI W. MEDL-Net: A model-based neural network for MRI reconstruction with enhanced deep learned regularizers[J]. Magn Reson Med, 2023, 89(5): 2062-2075.
[16]	BECK A, TEBOULLE M. A fast iterative shrinkage-thresholding algorithm for linear inverse problems[J]. SIAM J Imaging Sci, 2009, 2(1): 183-202.
[17]	ZHANG J, GHANEM B. ISTA-Net: Interpretable optimization-inspired deep network for image compressive sensing[C]// IEEE conference on computer vision and pattern recognition, Salt Lake City, USA: IEEE, 2018: 1828-1837.
[18]	YOU D, XIE J, ZHANG J. ISTA-NET++: Flexible deep unfolding network for compressive sensing[C]// IEEE International Conference on Multimedia and Expo, Shenzhen, China: IEEE, 2021: 1-6.
[19]	DUAN J, SCHLEMPER J, QIN C, et al. VS-Net: variable splitting network for accelerated parallel MRI reconstruction[C]// International Conference on Medical Image Computing and Computer-Assisted Intervention, Shenzhen, China: Springer, 2019: 713-722.
[20]	YE J C. Compressed sensing MRI: A review from signal processing perspective[J]. BMC Biomedical Engineering, 2019, 1(1): 1-17.
[21]	ARGHYA P, YOGESH R. A review and experimental evaluation of deep learning methods for MRI reconstruction[J]. J Mach Learn Biomed Imaging, 2022, 3: 1-55.
[22]	CHEN Q, SHAH N J, WORTHOFF W A. Compressed sensing in sodium magnetic resonance imaging: techniques, applications, and future prospects[J]. J Magn Reson Imaging, 2022, 55(5): 1340-1356.
[23]	JURE Z, FLORIAN K, ANUROOP S, et al. FastMRI: an open dataset and benchmarks for accelerated MRI[J]. arXiv preprint, arXiv: 1811.08839, 2019.
[24]	PENG L, MARK J, MICHAEL L, et al. ESPIRiT-an eigenvalue approach to autocalibrating parallel MRI: where SENSE meets GRAPPA[J]. Magn Reson Med, 2014, 71(3): 990-1001.
[25]	HONG G, WEI Y, MORLEY W, et al. Dual-domain accelerated MRI reconstruction using transformers with learning-based undersampling[J]. Comput Med Imag Graph, 2023, 106: 102206.

评价指标	AF	L1-ESPIRiT	ISTA	U-Net	ISTA-Net+	VS-Net	ISTAVS-Net
PSNR/dB	2	32.671	34.051	40.926	40.879	40.675	42.021
	4	28.868	30.390	36.911	36.281	36.478	37.724
	6	27.234	29.567	34.052	33.808	33.932	34.609
	8	26.976	28.446	33.287	32.713	33.184	33.341
SSIM	2	0.8802	0.9301	0.9573	0.9470	0.9547	0.9604
	4	0.8275	0.8680	0.9257	0.9128	0.9208	0.9311
	6	0.7608	0.8424	0.8950	0.8807	0.8936	0.9002
	8	0.7020	0.8205	0.8825	0.8679	0.8833	0.8839
NMSE	2	0.0205	0.0160	0.0107	0.0113	0.0109	0.0105
	4	0.0384	0.0362	0.0189	0.0193	0.0190	0.0183
	6	0.0478	0.0438	0.0277	0.0280	0.0275	0.0271
	8	0.0578	0.0567	0.0347	0.0350	0.0316	0.0304

	轴位			冠状位
	L1-ESPIRiT	ISTA	ISTAVS-Net	L1-ESPIRiT	ISTA	ISATVS-Net
PSNR/dB	31.743	34.051	38.952		27.234	29.567	34.609
SSIM	0.8180	0.9089	0.9382	0.7608	0.8424	0.9002
NMSE	0.0467	0.0395	0.0227	0.0478	0.0438	0.0271
Times/s	196.12	71.66	2.11	153.95	80.13	1.04