4759

High Resolution 3D Isotropic Multi-Contrast Brain Imaging using APIR4EMC
Chaoping Zhang1,2, Alexandra Cristobal-Huerta2, Juan Antonio Hernandez-Tamames2, Stefan Klein1,2, and Dirk H.J. Poot1,2

1Department of Medical Informatics, Erasmus MC, Rotterdam, Netherlands, 2Department of Radiology and Nuclear Medicine, Erasmus MC, Rotterdam, Netherlands

Synopsis

The long scan time of the brain MRI limits its applicability in high resolution 3D isotropic imaging. By using the recent Autocalibrated Parallel Imaging Reconstruction for Extended Multi-Contrast (APIR4EMC) method, we propose a high resolution (1 mm) 3D isotropic multi-contrast (T1, T1-Fatsat, T2, PD, FLAIR) brain imaging method with scan time around 10 min on a 3T MR scanner with an 8-channel brain coil. Experimental results demonstrate the effectiveness of this method.

Introduction

The long scan time of the brain MRI limits its applicability in high resolution 3D isotropic imaging. By using the recent Autocalibrated Parallel Imaging Reconstruction for Extended Multi-Contrast (APIR4EMC) [1] method, we propose a high resolution (1 mm) 3D isotropic multi-contrast (T1, T1-Fatsat, T2, PD, FLAIR) brain imaging method with scan time around 10 min on a 3T MR scanner with an 8-channel brain coil. Experimental results demonstrate the effectiveness of this method.

Methods

To improve the image quality for multi-contrast imaging from the conventional parallel imaging, APIR4EMC [1] was proposed by encoding the coil sensitivity and also the contrasts correlation into the autocalibration kernel in a GRAPPA-like reconstruction. Based on this, we further propose and validate this method for in-vivo brain imaging with the contrasts T1, T1-Fatsat, T2, PD, and FLAIR. The acquisition was performed on a 3T GE (MR750) scanner with an 8-channel brain coil, using the variable flip angle Fast Spin Echo (FSE) [2] sequence. Common settings: $matrix=224\times224\times178$, $FOV=224\times224\times178mm^3$, $BW/pixel=41.67kHz$, $echoes2skip=2$ and Radial trajectory [3] with k-space corners skipped. For T1-weighted contrast, $TR/TI/ESP/ETL=1062ms/500ms/5.360ms/62$. Fat saturation was additionally used for the acquisition of T1-Fatsat contrast. For T2/PD-weighted, $TR/ESP/ETL=2000ms/4.688ms/122$ with fast recovery. Except for the two skipped echoes, the first and the last 60 of the other echoes are separately placed in PD- and T2- weighted k-spaces, respectively (see Figure 1). For FLAIR, $TR/TI/ESP/ETL=5000ms/1700ms/5.360ms/120$. A k-space pattern with subsampling factor of $2\times3$ in two phase encoding (PE) dimensions with a $25\times25$ calibration region (see Figure 2) is used. The total effective scan time is 9 min.

The autocalibration kernel is learned from the ACS region of all contrasts. Because of the radial trajectory, the ACS region is filled by the first part of each echo train (ET). Differences in ET evolution among different contrasts (see Figure 1) can cause bias in the interpolation of unsampled positions using the kernel estimated from a small part of the ET. To compensate this, the ET of each contrast is stabilized by dividing each echo of each contrast by the norm of the corresponding echo in the additionally acquired zero phase encoding ET. To reduce Gibbs ringing and reduce noise amplification a Fermi filter ($kT=0.05$, $\mu=0.9$) is applied in each k-space before reconstruction.

For comparison we reconstruct each k-space separately with GRAPPA, using the same stabilization and Fermi filtering and with the same regularization strength (in kernel estimation) as used by APIR4EMC. The noise amplification map is computed for both APIR4EMC and GRAPPA with 50 pseudo replicas [4].

Results and Discussion

The reconstructed images and the noise amplification maps in axial plane are shown in Figure 3. The sagittal and coronal planes of the T1-Fatsat image are additionally shown in Figure 4. APIR4EMC reconstructed images show no obvious artifacts or excessive noise amplification. By additionally exploring the correlation among multi-contrast data, the noise amplification and artifacts are substantially reduced in APIR4EMC images, compared to the GRAPPA reconstruction.

Conclusion

Experimental results demonstrate that the APIR4EMC brain imaging method is able to acquire and reconstruct high resolution 3D isotropic brain images for multiple contrasts in around 10 min on a 3T MR scanner with an 8-channel brain coil.

Acknowledgements

No acknowledgement found.

References

1. Zhang C, Cristobal-Huerta A, Hernández Tamames J A, et al, APIR4EMC: Autocalibrated Parallel Imaging Reconstruction for Extended Multi-Contrast Imaging, Proceeding ISMRM, 2018, 4147.

2. Busse R F, Hariharan H, Vu A, et al. Fast spin echo sequences with very long echo trains: design of variable refocusing flip angle schedules and generation of clinical T2 contrast[J]. Magnetic Resonance in Medicine, 2006, 55(5): 1030-1037.

3. Busse R F, Brau A, Vu A, et al. Effects of refocusing flip angle modulation and view ordering in 3D fast spin echo[J]. Magnetic Resonance in Medicine, 2008, 60(3): 640-649.

4. Robson P M, Grant A K, Madhuranthakam A J, et al. Comprehensive quantification of signal-to-noise ratio and g-factor for image-based and k-space-based parallel imaging reconstructions[J]. Magnetic Resonance in Medicine, 2008, 60(4): 895-907.

Figures

Figure 1. Signal evolution for T1- , T1-Fatsat , PD- and T2-weighted, and FLAIR contrasts in the zero-phase-encoded echo train.

Figure 2. Illustration of the k-space pattern with subsampling factor $2 \times3$ in two PE dimensions. The k-space for each contrast has a matrix size of $224 \times224 \times 178$ in PE1, readout, and PE2 directions. The ACS region is $25\times 25$ in two PE directions. Color indicates echo number. Transparency indicates echo train number. Symbol indicates contrast type.

Figure 3. Reconstructed images (left) and noise amplification maps (right) for all contrasts in axial view by GRAPPA and APIR4EMC.

Figure 4. The sagittal and coronal planes of the T1-Fatsat image.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
4759