2440

A novel reconstruction method using regional constraints, designed for the dual-band EPI scanned with four-channel receiver coil elements
Hiroshi Toyoda1, Sosuke Yoshinaga2, Mitsuhiro Takeda2, and Hiroaki Terasawa2

1National Institute of Information and Communications Technology, Suita, Japan, 2Department of Structural Bioimaging, Kumamoto University, Kumamoto, Japan

Synopsis

A novel reconstruction method was proposed for dual-band EPI in an MRI system equipped with four-channel receiver coils. This method was based on a conventional kernel method utilizing an iterative calculation with regional constraints in the image domain. The method significantly improves the quality of the reconstructed images, even in the regions with less coil sensitivity. The results showed higher signal-to-noise ratio, less signal leakage, and better long-term stability in repetitions in comparison to the conventional method. The proposed method can be applied to clinical systems that have relatively few receiver coils, as well as animal systems.

INTRODUCTION

Simultaneous multi-slice (SMS) acquisition can be performed fairly well on a recent high-specification MRI system for humans with a redundant number of receiver coil elements. Thus, accurate slice separation for SMS data reconstruction continues to be challenging, especially when scanners equipped with relatively few receiver coil elements are used. In a system with relatively few receiver coil elements, it is difficult to implement SMS due to the gaps in signal sensitivity between the adjacent coil elements.

In this study, we propose a novel reconstruction method using regional constraints in the image domain for the dual-band EPI of in vivo normal rat brains scanned with an animal scanner with four-channel receiver coils. The method proposed in this study aims to reduce artifacts, such as Nyquist ghosts and inter-slice signal leakage, and to maximize and stabilize the signal-to-noise ratio (SNR) in reconstructed multi-band images.

METHODS

In vivo male Wister rat brains (n = 12) were scanned using a 7T animal scanner (Bruker, BioSpec70/20) equipped with a transceiver radiofrequency (RF) coil and a receiver array coil with four elements. A custom-built single-shot dual-band gradient echo 2D EPI sequence was used with blipped-controlled aliasing for the main scans, which were preceded by reference scans with a single-band 2D EPI sequence. Dual-band excitation pulses were designed on the basis of sinc functions with frequency offsets. A total of 78 slices (39 dual-band excitations) were imaged in a single repetition. The repetition time was 2,500 ms. The voxel size in the plane was 0.25 × 0.25 mm, and the slice thickness was 0.25 mm. The encoding and imaging matrix sizes in the read-out and phase-encoding directions were 96 and 56, respectively.

All the single-band and dual-band EPI raw data were processed using custom-built reconstruction software designed for the purpose of this study. The first step in the reconstruction was the slice separation from the folded images using a slice-GRAPPA method with 1x1-sized kernels in the k-space, which was followed by the EPI phase correction between the odd and even lines in the k-space. Then, slice separation was performed again using the optimal 3x3 size kernels on the phase-corrected data. Finally, to ensure better quality, regional constraints were introduced by masking the resulting images to correct the minor errors in the k-space data due to the relatively poor signal sensitivity. The workflow of the procedures is shown in Figure 1. The performance of the reconstruction methods was evaluated in terms of SNR, inter-slice leakage, and g-factor.

RESULTS

The proposed reconstruction method can provide accurate and stable results in reconstructed images for dual-band EPI scans obtained on a 7T animal scanner (equipped with a four-channel receiver coil). The reconstructed images using the proposed method were compared with those obtained using conventional kernel methods (Figure 2). Compared to the conventional method, the proposed method shows better image quality even in the region with insufficient signal sensitivity.

Figure 3 shows all the slices of the magnitude images reconstructed using the proposed method. Figure 4 shows the g-factor maps. The mean inter-slice leakage errors with the proposed method were remarkably improved compared to those of the conventional method (from 5.07 to 3.04), while the mean g-factor remained unchanged (from 0.60 to 0.61),

DISCUSSION

At the slice location where the coil sensitivity is low, the signals at the marginal portion in the k-space should be detected to be lower due to noise contamination. Therefore, due to the Fourier transform, erroneous signals are allocated to the voxels outside the body, which should not have a signal source. The results showed that the proposed method that imposes regional constraints in the image domain can used to modify the k-space data in order to eliminate the possibility of obtaining extracorporeal signal sources.

CONCLUSION

A novel reconstruction method was proposed to reduce artifacts and improve image quality in the dual-band EPI of in vivo normal rat brains. The proposed method allows slice acceleration in animal imaging by improving the reconstruction quality, which results in the ability to more quickly acquire whole-brain coverage of small animal brains without sacrificing spatial resolution in the slice-encoding direction (isotropic voxels can be achieved). Furthermore, the proposed method improves reconstruction accuracy in multi-band imaging; thus, it widens the range of applications of multi-band imaging, such as arterial spin labelling, functional, and diffusion MRI. The method can be used for slice acceleration in animal imaging; it can also be used to ensure more efficient multi-band EPI in humans, which can be scanned even with a clinical scanner system equipped with relatively few receiver coil elements.

Acknowledgements

This study was supported by JSPS KAKENHI Grants (Number 16K01970).

References

1) Setsompop K, gagoski BA, Polimeni JR, et al, Magn Reson Med 2012;67:1210-1224.

2) Cauley SF, Polimeni JR, Bhat H, et al, Magn Reson Med 2014;72:93-102.

3) Moeller S, Xu J, Auerback EJ, et al, In Proceedings of the 20th Annual meeting of ISMRM, Melbourne, Australia, 2012. p.519.

Figures

Figure 1. The flow chart of the procedures for the proposed method in iterative reconstruction with regional constraints in an image domain. The unfolded slice images for each slice were masked to eliminate the erroneous signals outside of the head regions .

Figure 2. Phase corrected magnitude images for dual-band EPI with CAIPI (FOV/2) shift after the slice unfolding procedures for the proposed method (A) and the conventional method (B). The images are shown as orthogonal planar views in the coronal, sagittal, and transversal planes. These were obtained from identical measurements. No averaging was used. The general image contrast does not seem to be significantly different. There seem to be regional gaps in signal intensity at the boundary between the adjacent coil sensitivity profiles in (B), however, the boundary gap not apparently seen in (A).

Figure 3. Magnitude images of the entire slices reconstructed using the proposed method. These images are displayed without any smoothing or averaging. A total of 78 slices (39 dual-band excitations) were imaged in a single repetition. The repetition time was 2,500 ms. The voxel size in the plane was 0.25 × 0.25 mm, and the slice thickness was 0.25 mm (gapless interleaved, isotropic voxel size). The encoding and imaging matrix sizes in the read-out and phase-encoding directions were 96 and 56, respectively. Neither in-plane acceleration nor partial Fourier sampling was used.

Figure 4. The g-factor maps of the entire slices calculated using the proposed method. The voxels with a higher g-factor are observed in the caudal parts in both slices-groups , which are located in the areas around the boundary between the sensitivity profiles of the two adjacent elements of the receiver array coil.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
2440