Only‐train‐once MR fingerprinting for B₀ and B₁ inhomogeneity correction in quantitative magnetization‐transfer contrast

Abstract

To develop a fast, deep-learning approach for quantitative magnetization-transfer contrast (MTC)-MR fingerprinting (MRF) that simultaneously estimates multiple tissue parameters and corrects the effects of B 0 and B 1 variations.Methods: An only-train-once recurrent neural network was designed to perform the fast tissue-parameter quantification for a large range of different MRF acquisition schedules. It enabled a dynamic scan-wise linear calibration of the scan parameters using the measured B 0 and B 1 maps, which allowed accurate, multiple-tissue parameter mapping. MRF images were acquired from 8 healthy volunteers at 3 T. Estimated parameter maps from the MRF images were used to synthesize the MTC reference signal (Z ref ) through Bloch equations at multiple saturation power levels. Results:The B 0 and B 1 errors in MR fingerprints, if not corrected, would impair the tissue quantification and subsequently corrupt the synthesized MTC reference images. Bloch equation-based numerical phantom studies and synthetic MRI analysis demonstrated that the proposed approach could correctly estimate water and semisolid macromolecule parameters, even with severe B 0 and B 1 inhomogeneities. Conclusion:The only-train-once deep-learning framework can improve the reconstruction accuracy of brain-tissue parameter maps and be further combined with any conventional MRF or CEST-MRF method.

Bloch simulator–driven deep recurrent neural network for magnetization transfer contrast MR fingerprinting and CEST imaging

“…The Bloch simulator needs to be retrained repeatedly for each MRF schedule tested, requiring intensive computation (about 27 h per epoch, including training data set generation and neural network training). However, the retraining of the Bloch simulator after each epoch during MRF optimization could be avoided if the RNN‐based Bloch simulator is trained with various tissue parameters and scan parameters using a recently developed Only‐Train‐Once MRF (OTOM) method 57,58 . The current deep Bloch simulator can be trained with a fixed duration of the MRF sequence (e.g., 40 dynamic scans), which is a limitation in the optimization of the number of dynamic scans for acquisition efficiency.…”

“…However, the retraining of the Bloch simulator after each epoch during MRF optimization could be avoided if the RNN-based Bloch simulator is trained with various tissue parameters and scan parameters using a recently developed Only-Train-Once MRF (OTOM) method. 57,58 The current deep Bloch simulator can be trained with a fixed duration of the MRF sequence (e.g., 40 dynamic scans), which is a limitation in the optimization of the number of dynamic scans for acquisition efficiency. However, the RNN-based OTOM framework can be trained only once and applied to different types of MRF schedule with various schedule lengths.…”

B1 inhomogeneity corrected CEST MRI based on direct saturation removed omega plot model at 5T

“…14 A number of retrospective correction methods have been developed to mitigate B 1 inhomogeneity induced artifact in CEST characterizations. [15][16][17][18][19] For example, Sun et al resolved the dependence of steady-state CEST contrast on B 1 field distribution and reduced the impact of B 1 inhomogeneity on CEST measurement by taking the induced labeling coefficient and spillover factor into account. 15 However, additional information (e.g., T 1 and T 2 ) was required.…”

Accelerating CEST imaging using a model‐based deep neural network with synthetic training data

“…Moreover, many early works trained and evaluated their networks on homogeneous data obtained from a few institutions, making them hard to generalize to various practical situations. 29,30 Following the successful applications of deep learning in a wide range of fields, learning-based approaches have received great interest in the CEST field, such as for rapid CEST MR fingerprinting, [31][32][33][34][35] B 0 or B 1 inhomogeneity correction, 36,37 and CEST data analysis or synthesis. [38][39][40][41][42] Specifically, for CEST image reconstruction, Guo et al 43 used a U-Net 44 to directly learn the mapping from undersampled source images to CEST contrast maps.…”

“…Following the successful applications of deep learning in a wide range of fields, learning‐based approaches have received great interest in the CEST field, such as for rapid CEST MR fingerprinting, 31–35 B 0 or B 1 inhomogeneity correction, 36,37 and CEST data analysis or synthesis 38–42 . Specifically, for CEST image reconstruction, Guo et al 43 used a U‐Net 44 to directly learn the mapping from undersampled source images to CEST contrast maps.…”