Tianyu Han^{1}, Teresa Nolte^{1,2}, Nicolas Gross-Weege^{1}, and Volkmar Schulz^{1}

To make the MRF technique most suitable for clinical needs, efforts are still to be made to accelerate MRF acquisitions while maintaining the accuracy in parameter determination. However, the dictionary calculation is a heavy computational burden for each trial MRF measurement within the optimization process. In this work, we present a numerical study on the optimization of MRF-FISP sequences by using a parallel tempering algorithm. Specifically, an optimization framework tailored for MRF with severe k-space undersampling was developed based on the previously proposed dictionary-free reconstruction (DFR). In vivo measurements were carried out to evaluate the performance of the optimized sequence.

This project has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement No. 667211.

1. Dan. M, et al. Magnetic Resonance Fingerprinting. Nature. 2013; 495: 187-193.

2. Tianyu. H, et al. Fast Dictionary-free Reconstruction in MR Fingerprinting. Proc. Intl. Soc. Mag. Reson. Med. 2018; 26: 2893.

3. Kendall A, An Introduction to Numerical Analysis (2nd edition), Section 3.8. John Wiley & Sons. New York,1988.

4. Ouri. C, et al. Algorithm comparison for schedule optimization in MR fingerprinting Magnetic Resonance Imaging. 2017; 41:15-21.

5. Matthias W, Extended phase graphs: Dephasing, RF pulses, and echoes ‐ pure and simple. Journal of Magnetic Resonance Imaging. 2015; 41: 266-295.

6. David. J. E, et al. Parallel tempering: Theory, applications, and new perspectives. Phys . Chem. Chem. Phys. 2005; 7: 3910-3916.

7. Yun. J, et al. MR fingerprinting using fast imaging with steady state precession (FISP) with spiral readout. Magn Reson Med. 2015; 74:1621-1631.

8. Karsten S, et al. Towards predicting the encoding capability of MR fingerprinting sequences. Magnetic Resonance Imaging. 2017; 41: 7-14.

The flowchart of the objective function calculation for the undersampling optimization. By performing the DFR, the matching error and dot-product profile can be estimated. Inner enlarged part: sampling points for the gradient calculation. The cross of two dash lines indicates the location of the maximum dot-product. In total eight sampling points are chosen in the periphery of the max dot-product entry.

(A): $$$P_{Fully}$$$ obtained from the proposed optimization algorithm. (B): Brain tissue relaxation parameters.

(A), (B): The $$$T_1$$$ fluctuation $$$\Delta T_1$$$ induced by undersampling in fully converged optimization, i.e., $$$P_{Fully}$$$, and Jiang's acquisitions, i.e., $$$P_{Jiang}$$$. (C), (D): The $$$T_2$$$ fluctuation $$$\Delta T_2$$$ induced by undersampling in $$$P_{Fully}$$$ and $$$P_{Jiang}$$$ acquisitions.

Top row: quantitative CSF maps reconstructed from the original $$$P_{Jiang}$$$ acquisition. Note the distinct difference marked by white arrows in the CSF distribution in $$$T_1$$$ and $$$T_2$$$ maps. Second row: CSF maps reconstructed from the fully optimized $$$P_{Fully}$$$ acquisition. Bottom row: CSF maps reconstructed from $$$P_{Early}$$$. Inconsistent maps (marked by white arrows in the last row) are observed in the undersampling acquisition.

Top row: quantitative GM and WM maps reconstructed from the original Jiang's acquisition. Second row: GM and WM maps reconstructed from the fully optimized acquisition. Bottom row: GM and WM maps reconstructed from the not fully optimized acquisition. Note the biased estimations (marked by white arrows) from the undersampling acquisition.