Free‐breathing 3D cardiac T₁mapping with transmit B₁correction at 3T

Abstract

Purpose: To develop a cardiac T 1 mapping method for free-breathing 3D T 1 mapping of the whole heart at 3 T with transmit B 1 (B + 1 ) correction. Methods: A free-breathing, electrocardiogram-gated inversion-recovery sequence with spoiled gradient-echo readout was developed and optimized for cardiac T 1 mapping at 3 T. High-frame-rate dynamic images were reconstructed from sparse (k,t)-space data acquired along a stack-of-stars trajectory using a subspace-based method for accelerated imaging. Joint T 1 and flip-angle estimation was performed in T 1 mapping to improve its robustness to B + 1 inhomogeneity. Subject-specific timing of data acquisition was used in the estimation to account for natural heart-rate variations during the imaging experiment.Results: Simulations showed that accuracy and precision of T 1 mapping can be improved with joint T 1 and flip-angle estimation and optimized electrocardiogramgated spoiled gradient echo-based inversion-recovery acquisition scheme. The phantom study showed good agreement between the T 1 maps from the proposed method and the reference method. Three-dimensional cardiac T 1 maps (40 slices) were obtained at a 1.9-mm in-plane and 4.5-mm through-plane spatial resolution from healthy subjects (n = 6) with an average imaging time of 14.2 ± 1.6 minutes (heartbeat rate: 64.2 ± 7.1 bpm), showing myocardial T 1 values comparable to those obtained from modified Look-Locker inversion recovery. The proposed method generated B + 1 maps with spatially smooth variation showing 21%-32% and 11%-15% variations across the septal-lateral and inferior-anterior regions of the myocardium in the left ventricle. Conclusion:The proposed method allows free-breathing 3D T 1 mapping of the whole heart with transmit B 1 correction in a practical imaging time.

Systematic review of reconstruction techniques for accelerated quantitative MRI

“…To include auxiliary parameters in the dictionary generation, the dictionary needs to be extended. Such extensions may include slice profile effects, 102,112,119,120,197,230 $$$ {\mathrm{B}}_1 $$$ inhomogeneity, 36,66,165,197,228‐230,241‐243 receiver phase, 230 $$$ {\mathrm{B}}_0 $$$ inhomogeneity, 40,233,244 relaxation effects during the inversion and preparation pulses, 120 and partial volume effects 227 …”

“…Regularizers can be introduced to add prior information, such as sparsity, 164 to the optimization problem defined in Equation (), yielding: $$$$ \hat{\boldsymbol{\sigma}}=\underset{\boldsymbol{\sigma}}{\arg\ \min }{\left\Vert \boldsymbol{Y}-\mathit{\mathcal{E}\psi}\boldsymbol{\sigma} \right\Vert}_2^2+\sum \limits_i{\lambda}_i{\left\Vert {\phi}_i\left(\boldsymbol{\sigma} \right)\right\Vert}_l. $$$$ Subspace constrained reconstruction has been used jointly with a spatially sparse representation of $$$ \boldsymbol{\sigma} $$$ using the wavelet transform, 110‐112,119,120,163 total variation, 117,119,123,124,129,143,145,155,162,165 finite difference, 164 Fourier representation of images in the temporal direction, 165 low‐rank priors, 112,115,126,128,135‐139,141,142,145,166‐172 and deep learning priors 173 …”

“…Most of the works use alternate direction method of multipliers with variable splitting for solving subspace constrained reconstruction problems. 112,115,116,118,121,122,126,128,130,[139][140][141][142]144,149,[162][163][164][165][166]169,183,187 Other solvers include the nonlinear conjugate gradient descent with backtracking line search, 120 the orthogonal matching pursuit method, 143,190 the linear conjugate gradient algorithm 123 and the fast iterative shrinkage-thresholding algorithm. 135…”

“…Subspace constrained reconstruction has been used jointly with a spatially sparse representation of 𝝈 using the wavelet transform, 110-112,119,120,163 total variation, 117,119,123,124,129,143,145,155,162,165 finite difference, 164 Fourier representation of images in the temporal direction, 165 low-rank priors, 112,115,126,128,[135][136][137][138][139]141,142,145,[166][167][168][169][170][171][172] and deep learning priors. 173 Alternative ways to exploit the low-rank nature of the signal.…”

Motion-compensated T1 mapping in cardiovascular magnetic resonance imaging: a technical review

“…Han et al used a non-Cartesian stack-of-stars trajectory to obtain free-breathing maps and maps with a mm resolution in an average scan time of 14.2 min ( 80 ). A combination of training data with limited -space coverage and a high sampling rate and imaging data with full -space coverage sparsely sampled across respiratory bins was used to track motion and collect imaging data.…”