Self-supervised Magnetic Resonance Fingerprint Parameter Quantization Method Based on Imaging Physical Model and Manifold Structure

doi:10.11938/cjmr20253173

Abstract

Abstract:

Magnetic resonance fingerprint (MRF) is an efficient multi-parameter quantitative imaging technology. However, traditional methods relying on signal dictionaries for parameter quantization are plagued by significant discretization errors and low matching efficiency. To overcome the limitations of existing supervised learning approaches that depend on pseudo-labels and lack physical interpretability, this study proposes a self-supervised parameter quantization method that integrates imaging physical models and manifold structure modeling. This method establishes reliable unlabeled constraints through Bloch equation-driven self-supervised physical consistency learning. By incorporating manifold structure-driven knowledge distillation, it transfers features of long frames to short frame models, realizing joint optimization of physical constraints and structural priors, thereby improving both accuracy and efficiency under unlabeled conditions. Experiments have verified this method’s superior accuracy and robustness, providing a novel approach for efficient and reliable MRF parameter estimation.

Key words: magnetic resonance fingerprint imaging, self-supervised learning, parameter quantization, imaging physical model, manifold structure

CLC Number:

LI Xiaodi, JI Yuping, HU Yue. Self-supervised Magnetic Resonance Fingerprint Parameter Quantization Method Based on Imaging Physical Model and Manifold Structure[J]. Chinese Journal of Magnetic Resonance, 2026, 43(1): 46-60.

Figures/Tables 15

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Fig. 5

Fig. 6

Table 1

Fig. 7

Fig. 8

Fig. 9

Table 2

Table 3

Table 4

Table 5

Fig. 10

References 31

[1]	MA D, GULANI V, SEIBERLICH N, et al. Magnetic resonance fingerprinting[J]. Nature, 2013, 495(7440): 187-192. doi: 10.1038/nature11971
[2]	HUANG M, LI S Y, CHEN J B, et al. Progress of magnetic resonance fingerprinting technology and its clinical application[J]. Chinese J Magn Reson, 2023, 40(2): 207-219.
	黄敏, 李思怡, 陈军波, 等. 磁共振指纹成像技术及临床应用的进展[J]. 波谱学杂志, 2023, 40(2): 207-219. doi: 10.11938/cjmr20223034
[3]	CAO X, LIAO C, ZHOU Z, et al. DTI-MR fingerprinting for rapid high-resolution whole-brain T₁, T₂, proton density, ADC, and fractional anisotropy mapping[J]. Magn Reson Med, 2024, 91(3): 987-1001. doi: 10.1002/mrm.v91.3
[4]	LENG Y J, WU Q W, LI Y X, et al. Quantitative detection of brain tissue parameters based on magnetic resonance fingerprinting and its preliminary application in brain diseases[J]. Chinese Journal of Clinical Neurosciences, 2020, 28(5): 579-583.
	冷一峻, 吴秋雯, 李郁欣, 等. 基于磁共振指纹技术的脑组织参数定量检测及在脑疾病中的初步应用[J]. 中国临床神经科学, 2020, 28(5): 579-583.
[5]	KRETZLER M E, HUANG S S, SUN J E P, et al. Free-breathing qRF-MRF with pilot tone respiratory motion navigator for T₁, T₂, T₂^*, and off-resonance mapping of the human body at 3 T[J]. Magn Reson Mater Phys, 2025, 38(1): 85-95. doi: 10.1007/s10334-024-01209-z
[6]	RATA M, ORTON M R, TUNARIU N, et al. Repeatability of quantitative MR fingerprinting for T₁ and T₂ measurements of metastatic bone in prostate cancer patients[J]. Eur Radiol, 2025, 35: 2487-2498. doi: 10.1007/s00330-024-11162-z
[7]	MCGIVNEY D F, PIERRE E, MA D, et al. SVD compression for magnetic resonance fingerprinting in the time domain[J]. IEEE Trans Med Imaging, 2014, 33(12): 2311-2322. doi: 10.1109/TMI.2014.2337321
[8]	ASSLANDER J, CLOOS M A, KNOLL F, et al. Low rank alternating direction method of multipliers reconstruction for MR fingerprinting[J]. Magn Reson Med, 2018, 79(1): 83-96. doi: 10.1002/mrm.26639 pmid: 28261851
[9]	CAULEY S F, SETSOMPOP K, MA D, et al. Fast group matching for MR fingerprinting reconstruction[J]. Magn Reson Med, 2015, 74(2): 523-528. doi: 10.1002/mrm.25439 pmid: 25168690
[10]	YANG M, MA D, JIANG Y, et al. Low rank approximation methods for MR fingerprinting with large scale dictionaries[J]. Magn Reson Med, 2018, 79(4): 2392-2400. doi: 10.1002/mrm.26867 pmid: 28804918
[11]	WANG Z, ZHANG J, CUI D, et al. Magnetic resonance fingerprinting using a fast dictionary searching algorithm: MRF-ZOOM[J]. IEEE Trans Biomed Eng, 2018, 66(6): 1526-1535. doi: 10.1109/TBME.10
[12]	NALLAPAREDDY N, RAY S. Inferring multiple tissue properties from magnetic resonance fingerprinting images[C]// Proceedings of the AAAI Conference on Artificial Intelligence. Palo Atto, CA: AAAI Press, 2022, 36(11): 12587-12593.
[13]	VIRTUE P, STELLA X Y, LUSTIG M. Better than real: Complex-valued neural nets for MRI fingerprinting[C]// IEEE International Conference on Image Processing (ICIP). Beijing, China: IEEE, 2017: 3953-3957.
[14]	OKSUZ I, CRUZ G, CLOUGH J, et al. Magnetic resonance fingerprinting using recurrent neural networks[C]// IEEE International Symposium on Biomedical Imaging (ISBI). Venice, Italy: IEEE, 2019: 1537-1540.
[15]	BALSIGER F, JUNGO A, SCHEIDEGGER O, et al. Spatially regularized parametric map reconstruction for fast magnetic resonance fingerprinting[J]. Med Image Anal, 2020, 64: 101741. doi: 10.1016/j.media.2020.101741
[16]	FANG Z, CHEN Y, LIU M, et al. Deep learning for fast and spatially constrained tissue quantification from highly accelerated data in magnetic resonance fingerprinting[J]. IEEE Trans Med Imaging, 2019, 38(10): 2364-2374. doi: 10.1109/TMI.42
[17]	FANG Z, CHEN Y, NIE D, et al. Rca-u-net: Residual channel attention u-net for fast tissue quantification in magnetic resonance fingerprinting[C]// Medical Image Computing and Computer Assisted Intervention-MICCAI International Conference. Shenzhen, China. Cham: Springer, 2019: 101-109.
[18]	LI P, HU Y. Deep magnetic resonance fingerprinting based on local and global vision transformer[J]. Med Image Anal, 2024, 95: 103198. doi: 10.1016/j.media.2024.103198
[19]	FATANIA K, CHAU K Y, PIRKL C M, et al. Nonlinear equivariant imaging: Learning multi-parametric tissue mapping without ground truth for compressive quantitative mri[C]// International Symposium on Biomedical Imaging (ISBI). Cartagena, Colombia: IEEE, 2023: 1-4.
[20]	HAMILTON J I. A self-supervised deep learning reconstruction for shortening the breathhold and acquisition window in cardiac magnetic resonance fingerprinting[J]. Front Cardiovasc Med, 2022, 9: 928546. doi: 10.3389/fcvm.2022.928546
[21]	MAYO P, CENCINI M, FATANIA K, et al. Deep image priors for magnetic resonance fingerprinting with pretrained Bloch-consistent denoising autoencoders[C]// IEEE International Symposium on Biomedical Imaging (ISBI). Athens, Greece: IEEE, 2024: 1-4.
[22]	ULYANOV D, VEDALDI A, LEMPITSKY V. Deep image prior[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Salt Lake City, UT, USA: IEEE, 2018: 9446-9454.
[23]	YUAN W, GU X, DAI Z, et al. Neural window fully-connected crfs for monocular depth estimation[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans, LA, USA: IEEE, 2022: 3916-3925.
[24]	RONNEBERGER O, FISCHER P, BROX T. U-net: Convolutional networks for biomedical image segmentation[C]// Medical Image Computing and Computer-Assisted Intervention-MICCAI International Conference. Munich, Germany. Cham: Springer, 2015: 234-241.
[25]	HU Y, LI P, CHEN H, et al. High-quality MR fingerprinting reconstruction using structured low-rank matrix completion and subspace projection[J]. IEEE Trans Med Imaging, 2021, 41(5): 1150-1164. doi: 10.1109/TMI.2021.3133329
[26]	COLLINS D L, ZIJDENBOS A P, KOLLOKIAN V, et al. Design and construction of a realistic digital brain phantom[J]. IEEE Trans Med Imaging, 2002, 17(3): 463-468. doi: 10.1109/42.712135
[27]	AUBERT-BROCHE B, EVANS A C, COLLINS L. A new improved version of the realistic digital brain phantom[J]. NeuroImage, 2006, 32(1): 138-145. doi: 10.1016/j.neuroimage.2006.03.052
[28]	WEIGEL M. Extended phase graphs: dephasing, RF pulses, and echoes-pure and simple[J]. Magn Reson Imaging, 2015, 41(2): 266-295.
[29]	SOYAK R, NAVRUZ E, ERSOY E O, et al. Channel attention networks for robust MR fingerprint matching[J]. IEEE Trans Biomed Eng, 2021, 69(4): 1398-1405. doi: 10.1109/TBME.2021.3116877
[30]	CHEN D, TACHELLA J, DAVIES M E. Equivariant imaging: Learning beyond the range space[C]// Proceedings of the IEEE/CVF International Conference on Computer Vision. Montreal, QC, Canada: IEEE, 2021: 4379-4388.
[31]	BIPIN MEHTA B, COPPO S, FRANCES MCGIVNEY D, et al. Magnetic resonance fingerprinting: a technical review[J]. Magn Reson Med, 2019, 81(1): 25-46. doi: 10.1002/mrm.27403 pmid: 30277265

方法	T₁			T₂
方法	NMSE	SSIM	PSNR	NMSE	SSIM	PSNR
SCQ^[16]	0.0071±0.0006	0.9809±0.0044	30.91±0.7194	0.0264±0.0039	0.9345±0.0168	25.37±0.7681
SPR^[15]	0.0081±0.0007	0.9800±0.0043	30.32±0.4961	0.0181±0.0046	0.9526±0.0147	27.24±1.1641
CONV-ICA^[29]	0.0060±0.0005	0.9842±0.0029	31.63±0.5046	0.0179±0.0025	0.8704±0.0502	27.21±0.3463
EI^[30]	0.0144±0.0058	0.9783±0.0036	28.15±1.7734	0.0385±0.0088	0.9019±0.0168	22.19±2.1771
NLEI^[19]	0.0147±0.0038	0.9755±0.0036	27.87±1.0816	0.0184±0.0016	0.9725±0.0031	27.06±0.4581
Distilled SPQ （本文方法）	0.0099±0.0030	0.9800±0.0030	29.60±1.2653	0.0172±0.0021	0.9003±0.0512	27.54±0.3693

方法	T₁/ms			T₂/ms
方法	白质	灰质	脑脊液	白质	灰质	脑脊液
参考值	794.6±58.6	1346.2±78.5	3357.9±102.3	73.9±5.5	90.4±5.3	820.7±111.3
Distilled SPQ（本文方法）	812.1±55.5	1373.0±87.3	3222.7±174.9	71.6±7.8	84.8±8.9	799.1±157.4
文献[31]	788~898	1286~1393	/	63~80	78~117	/

方法	T₁ NMSE	T₂ NMSE
SPQ-200	0.0227±0.0028	0.0208±0.0026
SPQ-500	0.0092±0.0029	0.0194±0.0023
Distilled SPQ（本文方法）	0.0099±0.0030	0.0172±0.0021

λ	T₁ NMSE	T₂ NMSE
0.01	0.0104±0.0032	0.0185±0.0024
0.1	0.0102±0.0031	0.0180±0.0023
1.0	0.0099±0.0030	0.0172±0.0021
10	0.0109±0.0034	0.0200±0.0025

方法	训练时间	参数量化时间
SCQ^[16]	6.5 h	0.173 s
SPR^[15]	6.2 h	0.172 s
CONV-ICA^[29]	6.0 h	0.170 s
EI^[30]	35.6 h	78.077 s
NLEI^[19]	49.7 h	1.204 s
Distilled SPQ（本文方法）	18.5 h	0.352 s