Regularized SUPER-CAIPIRINHA: Accelerating 3D variable flip-angle T1 mapping with accurate and efficient reconstruction
Fan Yang
Institute for Medical Imaging Technology, School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, People's Republic of China
Search for more papers by this authorJian Zhang
United Imaging Healthcare Co., Ltd, Shanghai, People's Republic of China
Search for more papers by this authorGuobin Li
United Imaging Healthcare Co., Ltd, Shanghai, People's Republic of China
Search for more papers by this authorJiayu Zhu
United Imaging Healthcare Co., Ltd, Shanghai, People's Republic of China
Search for more papers by this authorXin Tang
United Imaging Healthcare Co., Ltd, Shanghai, People's Republic of China
Institute for Medical Imaging Technology, School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, People's Republic of China
Search for more papers by this authorCorresponding Author
Chenxi Hu
Institute for Medical Imaging Technology, School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, People's Republic of China
Correspondence
Chenxi Hu, Institute for Medical Imaging Technology, School of Biomedical Engineering, Shanghai Jiao Tong University, 415 S Med-X Center, 1954 Huashan Road, Shanghai 200030, China.
Email: [email protected]
Search for more papers by this authorFan Yang
Institute for Medical Imaging Technology, School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, People's Republic of China
Search for more papers by this authorJian Zhang
United Imaging Healthcare Co., Ltd, Shanghai, People's Republic of China
Search for more papers by this authorGuobin Li
United Imaging Healthcare Co., Ltd, Shanghai, People's Republic of China
Search for more papers by this authorJiayu Zhu
United Imaging Healthcare Co., Ltd, Shanghai, People's Republic of China
Search for more papers by this authorXin Tang
United Imaging Healthcare Co., Ltd, Shanghai, People's Republic of China
Institute for Medical Imaging Technology, School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, People's Republic of China
Search for more papers by this authorCorresponding Author
Chenxi Hu
Institute for Medical Imaging Technology, School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, People's Republic of China
Correspondence
Chenxi Hu, Institute for Medical Imaging Technology, School of Biomedical Engineering, Shanghai Jiao Tong University, 415 S Med-X Center, 1954 Huashan Road, Shanghai 200030, China.
Email: [email protected]
Search for more papers by this authorAbstract
Purpose
To propose an acceleration method for 3D variable flip-angle (VFA) T1 mapping based on a technique called shift undersampling improves parametric mapping efficiency and resolution (SUPER).
Methods
The proposed method incorporates strategies of SUPER, controlled aliasing in volumetric parallel imaging (CAIPIRINHA), and total variation-based regularization to accelerate 3D VFA T1 mapping. The k-space sampling grid of CAIPIRINHA is internally undersampled with SUPER along the contrast dimension. A proximal algorithm was developed to preserve the computational efficiency of SUPER in the presence of regularization. The regularized SUPER-CAIPIRINHA (rSUPER-CAIPIRINHA) was compared with low rank plus sparsity (L + S), reconstruction of principal component coefficient maps (REPCOM), and other SUPER-based methods via simulations and in vivo brain T1 mapping. The results were assessed quantitatively with NRMSE and structural similarity index measure (SSIM), and qualitatively by two experienced reviewers.
Results
rSUPER-CAIPIRINHA achieved a lower NRMSE and higher SSIM than L + S (0.11 ± 0.01 vs. 0.19 ± 0.03, p < 0.001; 0.66 ± 0.05 vs. 0.37 ± 0.03, p < 0.001) and REPCOM (0.16 ± 0.02, p < 0.001; 0.46 ± 0.04, p < 0.001). The reconstruction time of rSUPER-CAIPIRINHA was 6% of L + S and 2% of REPCOM. For the qualitative comparison, rSUPER-CAIPIRINHA showed improvement of overall image quality and reductions of artifacts and blurring, although with a lower apparent SNR. Compared with 2D SUPER-SENSE, rSUPER-CAIPIRINHA significantly reduced NRMSE (0.11 ± 0.01 vs. 0.23 ± 0.04, p < 0.001) and generated less noisy reconstructions.
Conclusion
By incorporating SUPER, CAIPIRINHA, and regularization, rSUPER-CAIPIRINHA mitigated noise amplification, reduced artifacts and blurring, and achieved faster reconstructions compared with L + S and REPCOM. These advantages render 3D rSUPER-CAIPIRINHA VFA T1 mapping potentially useful for clinical applications.
CONFLICT OF INTEREST
Jian Zhang, Guobin Li, Jiayu Zhu, and Xin Tang are employees of United Imaging Healthcare Co., Ltd, Shanghai, China.
Supporting Information
| Filename | Description |
|---|---|
| mrm29714-sup-0001-SupInfo.pdfPDF document, 2.2 MB |
FIGURE S1. The pseudorandom undersampling pattern of L + S and REPCOM at R = 16. White dots represent sampled points in k-space. FIGURE S2. The g-factor map, d-factor maps and their overlays over an anatomical image for 4-fold CAIPIRINHA (Column 1) and 16-fold SUPER-CAIPIRINHA (Columns 2 and 3). The g-factor map describes noise amplification in the raw image with parallel imaging. The d-factor maps describe noise amplification in the parametric maps due to both parallel imaging and contrast-domain accelerations. The regional alignment of the hot areas between the g-factor and d-factor maps indicates the noise amplification in the T1 and spin density maps is mainly caused by insufficient coil coverage. FIGURE S3. The Bland–Altman analysis comparing non-acceleration gold standard with rSUPER-CAIPIRINHA (A), L + S (B), and REPCOM (C) in the gray matter (square) and white matter (circle) ROIs. rSUPER-CAIPIRINHA resulted in considerably less biases compared with the other two methods. FIGURE S4. Reconstruction results of rSUPER-CAIPIRINHA, REPCOM and L + S at R = 16 with the SUPER-CAIPIRINHA sampling pattern. L + S failed the reconstruction since it is model-free and thus needs pseudorandom sampling to create incoherent artifacts. REPCOM led to a similar reconstruction quality with rSUPER-CAIPIRINHA, since both of them are model-based. However, notice that REPCOM needs 350 min for the reconstruction while rSUPER-CAIPIRINHA only needs 5 min. FIGURE S5. Results of Monte-Carlo simulations for rSUPER-CAIPIRINHA (Row 1), REPCOM (Row 2) and L + S (Row 3) at an acceleration rate of 16. One hundred sets of 3D VFA data were synthesized based on the gold standard T1 and spin density maps in a subject and simulations of i.i.d. complex-valued Gaussian noise in k-space. Since REPCOM and L + S reconstructions are time-consuming, only a single y-z slice was reconstructed for comparison. Column 1 shows the mean T1 maps. Column 2 shows the bias maps. Column 3 shows the standard deviation maps. rSUPER-CAIPIRINHA led to a considerablylower bias but a mildly higher standard deviation compared with L + S and REPCOM. |
| mrm29714-sup-0002-VideoS1.mp4MPEG-4 video, 796.5 KB | VIDEO S1. Reconstructed 3D T1 and spin density maps with NON-ACC (R = 1), CAIPIRINHA (R = 4) and rSUPER-CAIPIRINHA (R = 8 and R = 16). Retrospective undersampling was performed. Row 1 shows T1 maps and Row 2 shows spin density maps. |
| mrm29714-sup-0003-VideoS2.mp4MPEG-4 video, 809.3 KB | VIDEO S2. Reconstructed 3D T1 and spin density maps with prospective undersampling. rSUPER-CAIPIRINHA reduced the 3D scan time from 14:10 min to 0:59 min with 16-fold acceleration. The reconstruction quality was consistent with that of retrospective undersampling. |
| mrm29714-sup-0004-VideoS3.mp4MPEG-4 video, 3.2 MB | VIDEO S3. Reconstructed 3D isotropic-resolution (1.6 × 1.6 × 1.6 mm3) T1 and spin density maps of the whole upper brain for 1 healthy subject. The scan time of CAIPIRINHA and rSUPER-CAIPIRINHA was 11:42 min and 2:54 min, respectively. |
Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
REFERENCES
- 1Deoni SCL, Peters TM, Rutt BK. High-resolution T1 and T2 mapping of the brain in a clinically acceptable time with DESPOT1 and DESPOT2. Magn Reson Med. 2005; 53: 237-241.
- 2Vrenken H, Geurts JJG, Knol DL, et al. Whole-brain T1 mapping in multiple sclerosis: global changes of normal-appearing gray and white matter. Radiology. 2006; 240: 811-820.
- 3Koh TS, Thng CH, Lee PS, et al. Hepatic metastases: in vivo assessment of perfusion parameters at dynamic contrast-enhanced MR imaging with dual-input two-compartment tracer kinetics model. Radiology. 2008; 249: 307-320.
- 4Li Z, Sun J, Hu X, et al. Assessment of liver fibrosis by variable flip angle T1 mapping at 3.0T. J Magn Reson Imaging. 2016; 43: 698-703.
- 5Vymazal J, Righini A, Brooks RA, et al. T1 and T2 in the brain of healthy subjects, patients with Parkinson disease, and patients with multiple system atrophy: relation to iron content. Radiology. 1999; 211: 489-495.
- 6Zhu Z, Lebel RM, Bliesener Y, Acharya J, Frayne R, Nayak KS. Sparse precontrast T1 mapping for high-resolution whole-brain DCE-MRI. Magn Reson Med. 2021; 86: 2234-2249.
- 7Brookes JA, Redpath TW, Gilbert FJ, Murray AD, Staff RT. Accuracy of T1 measurement in dynamic contrast-enhanced breast MRI using two- and three-dimensional variable flip angle fast low-angle shot. J Magn Reson Imaging. 1999; 9: 163-171.
10.1002/(SICI)1522-2586(199902)9:2<163::AID-JMRI3>3.0.CO;2-L CAS PubMed Web of Science® Google Scholar
- 8Clique H, Cheng H-LM, Marie P-Y, Felblinger J, Beaumont M. 3D myocardial T1 mapping at 3T using variable flip angle method: pilot study. Magn Reson Med. 2014; 71: 823-829.
- 9Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. SENSE: sensitivity encoding for fast MRI. Magn Reson Med. 1999; 42: 952-962.
10.1002/(SICI)1522-2594(199911)42:5<952::AID-MRM16>3.0.CO;2-S CAS PubMed Web of Science® Google Scholar
- 10Griswold MA, Jakob PM, Heidemann RM, et al. Generalized autocalibrating partially parallel acquisitions (GRAPPA). Magn Reson Med. 2002; 47: 1202-1210.
- 11Breuer FA, Blaimer M, Mueller MF, et al. Controlled aliasing in volumetric parallel imaging (2D CAIPIRINHA). Magn Reson Med. 2006; 55: 549-556.
- 12Block KT, Uecker M, Frahm J. Model-based iterative reconstruction for radial fast spin-echo MRI. IEEE Trans Med Imaging. 2009; 28: 1759-1769.
- 13Wang X, Rosenzweig S, Scholand N, Holme HCM, Uecker M. Model-based reconstruction for simultaneous multi-slice mapping using single-shot inversion-recovery radial FLASH. Magn Reson Med. 2021; 85: 1258-1271.
- 14Hu C, Peters DC. SUPER: a blockwise curve-fitting method for accelerating MR parametric mapping with fast reconstruction. Magn Reson Med. 2019; 81: 3515-3529.
- 15Sumpf TJ, Uecker M, Boretius S, Frahm J. Model-based nonlinear inverse reconstruction for T2 mapping using highly undersampled spin-echo MRI. J Magn Reson Imaging. 2011; 34: 420-428.
- 16Zhao B, Lu W, Hitchens TK, Lam F, Ho C, Liang Z-P. Accelerated MR parameter mapping with low-rank and sparsity constraints. Magn Reson Med. 2015; 74: 489-498.
- 17Hu C, Reeves S, Peters DC, Twieg D. An efficient reconstruction algorithm based on the alternating direction method of multipliers for joint estimation of R2* and off-resonance in fMRI. IEEE Trans Med Imaging. 2017; 36: 1326-1336.
- 18Huang C, Graff CG, Clarkson EW, Bilgin A, Altbach MI. T2 mapping from highly undersampled data by reconstruction of principal component coefficient maps using compressed sensing. Magn Reson Med. 2012; 67: 1355-1366.
- 19Cruz G, Qi H, Jaubert O, et al. Generalized low-rank nonrigid motion-corrected reconstruction for MR fingerprinting. Magn Reson Med. 2022; 87: 746-763.
- 20Velikina JV, Alexander AL, Samsonov A. Accelerating MR parameter mapping using sparsity-promoting regularization in parametric dimension. Magn Reson Med. 2013; 70: 1263-1273.
- 21Doneva M, Börnert P, Eggers H, Stehning C, Sénégas J, Mertins A. Compressed sensing reconstruction for magnetic resonance parameter mapping. Magn Reson Med. 2010; 64: 1114-1120.
- 22Zhang T, Pauly JM, Levesque IR. Accelerating parameter mapping with a locally low rank constraint. Magn Reson Med. 2015; 73: 655-661.
- 23Christodoulou AG, Shaw JL, Nguyen C, et al. Magnetic resonance multitasking for motion-resolved quantitative cardiovascular imaging. Nat Biomed Eng. 2018; 2: 215-226.
- 24Zhu Y, Liu Y, Ying L, et al. SCOPE: signal compensation for low-rank plus sparse matrix decomposition for fast parameter mapping. Phys Med Biol. 2018; 63:185009.
- 25Zibetti MVW, Sharafi A, Otazo R, Regatte RR. Compressed sensing acceleration of biexponential 3D-T1ρ relaxation mapping of knee cartilage. Magn Reson Med. 2019; 81: 863-880.
- 26Wang X, Roeloffs V, Klosowski J, et al. Model-based T1 mapping with sparsity constraints using single-shot inversion-recovery radial FLASH. Magn Reson Med. 2018; 79: 730-740.
- 27Zhao B, Lam F, Liang Z. Model-based MR parameter mapping with sparsity constraints: parameter estimation and performance bounds. IEEE Trans Med Imaging. 2014; 33: 1832-1844.
- 28Meng Z, Guo R, Li Y, et al. Accelerating T2 mapping of the brain by integrating deep learning priors with low-rank and sparse modeling. Magn Reson Med. 2021; 85: 1455-1467.
- 29Li Y, Wang Y, Qi H, et al. Deep learning–enhanced T1 mapping with spatial-temporal and physical constraint. Magn Reson Med. 2021; 86: 1647-1661.
- 30Hamilton JI, Currey D, Rajagopalan S, Seiberlich N. Deep learning reconstruction for cardiac magnetic resonance fingerprinting T1 and T2 mapping. Magn Reson Med. 2021; 85: 2127-2135.
- 31Wu D, Kim K, Li Q. Computationally efficient deep neural network for computed tomography image reconstruction. Med Phys. 2019; 46: 4763-4776.
- 32Wang K, Kellman M, Sandino CM, et al. Memory-efficient learning for high-dimensional MRI reconstruction. In: M Bruijne, PC Cattin, S Cotin, et al., eds. Springer International Publishing; 2021: 461-470.
- 33Zeng G, Guo Y, Zhan J, et al. A review on deep learning MRI reconstruction without fully sampled k-space. BMC Med Imaging. 2021; 21: 195.
- 34Sidky EY, Lorente I, Brankov JG, Pan X. Do CNNs solve the CT inverse problem? IEEE Trans Biomed Eng. 2021; 68: 1799-1810.
- 35Knoll F, Murrell T, Sriram A, et al. Advancing machine learning for MR image reconstruction with an open competition: overview of the 2019 fastMRI challenge. Magn Reson Med. 2020; 84: 3054-3070.
- 36Antun V, Renna F, Poon C, Adcock B, Hansen AC. On instabilities of deep learning in image reconstruction and the potential costs of AI. Proc Natl Acad Sci U S A. 2020; 117: 30088-30095.
- 37Rudin LI, Osher S, Fatemi E. Nonlinear total variation based noise removal algorithms. Physica D. 1992; 60: 259-268.
- 38Goldstein T, Osher S. The split Bregman method for L1-regularized problems. SIAM J Imaging Sci. 2009; 2: 323-343.
- 39Otazo R, Candès E, Sodickson DK. Low-rank plus sparse matrix decomposition for accelerated dynamic MRI with separation of background and dynamic components. Magn Reson Med. 2015; 73: 1125-1136.
- 40Deoni SCL, Rutt BK, Peters TM. Rapid combined T1 and T2 mapping using gradient recalled acquisition in the steady state. Magn Reson Med. 2003; 49: 515-526.
- 41Parikh N, Boyd S. Proximal algorithms. Found Trends Optim. 2014; 1: 127-239.
10.1561/2400000003 Google Scholar
- 42Polson NG, Scott JG, Willard BT. Proximal algorithms in statistics and machine learning. Stat Sci. 2015; 30: 559-581.
- 43McKenzie CA, Yeh EN, Ohliger MA, Price MD, Sodickson DK. Self-calibrating parallel imaging with automatic coil sensitivity extraction. Magn Reson Med. 2002; 47: 529-538.
- 44Manuel A, Li W, Jellus V, Hughes T, Prasad PV. Variable flip angle-based fast three-dimensional T1 mapping for delayed gadolinium-enhanced MRI of cartilage of the knee: need for B1 correction. Magn Reson Med. 2011; 65: 1377-1383.
- 45Boudreau M, Tardif CL, Stikov N, Sled JG, Lee W, Pike GB. B1 mapping for bias-correction in quantitative T1 imaging of the brain at 3T using standard pulse sequences. J Magn Reson Imaging. 2017; 46: 1673-1682.
- 46Andreisek G, White LM, Yang Y, Robinson E, Cheng H-LM, Sussman MS. Delayed gadolinium-enhanced MR imaging of articular cartilage: three-dimensional T1 mapping with variable flip angles and B1 correction. Radiology. 2009; 252: 865-873.
- 47Nehrke K, Börnert P. DREAM—a novel approach for robust, ultrafast, multislice B1 mapping. Magn Reson Med. 2012; 68: 1517-1526.
- 48Wansapura JP, Holland SK, Dunn RS, Ball WS Jr. NMR relaxation times in the human brain at 3.0 Tesla. J Magn Reson Imaging. 1999; 9: 531-538.
10.1002/(SICI)1522-2586(199904)9:4<531::AID-JMRI4>3.0.CO;2-L CAS PubMed Web of Science® Google Scholar
- 49Weiger M, Pruessmann KP, Boesiger P. 2D SENSE for faster 3D MRI. Magn Reson Mater Phys Biol Med. 2002; 14: 10-19.
- 50Menon RG, Zibetti MVW, Jain R, Ge Y, Regatte RR. Performance comparison of compressed sensing algorithms for acceleratingT1ρ mapping of human brain. J Magn Reson Imaging. 2021; 53: 1130-1139.
- 51Hu C, Peters DC. High-resolution single-breath-hold cardiac T2 mapping based on SUPER-SENSE-PCA. Proceedings of SCMR 22nd Annual Scientific Sessions (SCMR 2019); 2019: 1017-1020. https://orcid.org/0000-0001-6969-8257.
