3947

Improving k-space physiological noise correction with motion correction in fMRI studies
Chisato Suzuki1, R. Allen Waggoner2, Kenji Haruhana2, Keiji Tanaka1,2, and Kenichi Ueno1

1fMRI support unit, RIKEN Center for Brain Science, Wako-shi, Japan, 2Laboratory for Cognitive Brain Mapping, RIKEN Center for Brain Science, Wako-shi, Japan

Synopsis

Although physiological noise correction could improve the signal quality, the signal contamination caused by subject motion would affect the ability to remove the noise. This study aims to characterize the impact of motion on the physiological noise correction in k-space and improve k-space physiological noise correction performance. The strength of motion significantly influenced the average tSNR after k-space physiological noise correction. By removing motion effect in advance, it is possible to make k-space physiological noise correction more effective.

INTRODUCTION

In fMRI studies, physiological noise such as heart beat and respiration could lower the signal quality and contribute to worsening detectability of neuronal responses. To remove such noise, many physiological noise correction methods have been developed for k-space data as well as image-space data1,2,3,4. In spite of more accurate synchronization of physiological changes with the time of data acquisition in k-space data, a previous study5 reported that k-space correction is not as effective as image-space correction. Although physiological noise was corrected at each k-space point for the k-space data and at each voxel for the image-space data, the subject motion could make the signal at each k-space point or each voxel contain a mixture of signals acquired at different time or spatial points. This contamination of the signal might cause the physiological noise correction to be less effective. A previous study6 reported that motion correction before physiological noise correction for image-space data decreased temporal standard deviation. Since noise is a global effect rather than local in k-space, motion could be more problematic for k-space data than that for image-space data. However, there have been no studies about the effect of motion on physiological noise correction in k-space. Our goal in this study is to evaluate motion effects on physiological noise correction and improve the performance of k-space physiological noise correction by removing it.

METHODS

Two EPI time series were acquired for each of the four healthy volunteers (mean age 34, age range 25-39, two females) with a 3x3x3 mm3 voxel size using a 3 Tesla whole-body MRI scanner with a 64-channel phased-array receive coil (Siemens Healthineers). Pacing visual stimuli were used to make the subjects perform a finger tapping task during scan. The scan parameters were as follows: TR/TE=2sec/25ms, FOV=192x192mm2, acceleration factor=3, 190 volumes, 43 slices to cover the whole brain. The subject’s heartbeat and respiration were recorded with a pulse oximeter and a pressure sensor placed on the abdominal region. To evaluate motion effect on k-space physiological noise correction, we compared the physiological noise corrected data with k-space motion correction beforehand (WithKmc) and with image-space motion correction afterwards (NoKmc). For WithKmc data, the motion was corrected in k-space using the six rigid motion parameters (three rotations and three shifts) of each volume calculated from reconstructed image data by using 3dvolreg in AFNI7. Then both cardiac and respiratory noise were removed from k-space data using a retrospective correction method similar to RETROKCOR1. To evaluate effects of motion on physiological noise correction, we determined how much the temporal signal to noise (tSNR) improved after physiological noise correction and examined the correlation between the improvements of tSNR average and the strength of motion.

RESULTS AND DISCUSSION

While the amount of improvement in tSNR average (ΔtSNR) by k-space physiological noise correction without motion correction was significantly correlated with the total variation of all three rotations, IS shift, and LR shift, there was no correlation between ΔtSNR of the data with k-space motion correction and the total variation of any of the motion parameters (Fig 1). Fig 2 also shows that the number of the voxels showing decreased tSNR after physiological noise correction reduced remarkably by applying k-space motion correction regardless of the strength of the motion. Moreover, the average of t-value of the regions showing activation significantly correlated with the tapping task was better for the data with k-space motion correction than the data without it. These results imply that as motion is more detrimental to the k-space correction of physiological noise, motion correction beforehand could improve performance of physiological noise correction in k-space.

CONCLUSION

Although the appropriate physiological noise correction could improve signal quality, it doesn’t work properly if other sources of noise such as motion were not taken into account in the correction. Removing this motion effect by applying k-space motion correction in advance could make k-space physiological noise correction more effective.

Acknowledgements

The study was supported by AMED under Grant Number JP18dm0207001 and by JSPS KAKENHI Grant Number JP16H06564

References

1. Hu X, Le TH, Parrish T, Erhard P. Retrospective estimation and correction of physiological fluctuation in functional MRI. Magn Reson Med. 1995 Aug;34(2):201-12.

2. Le TH, Hu X. Retrospective estimation and correction of physiological artifacts in fMRI by direct extraction of physiological activity from MR data. Magn Reson Med. 1996 Mar;35(3);290-8.

3. Wowk B, Mclntyre MC, Saunders JK. k-space detection and correction of physiological artifacts in fMRI. Magn Reson Med. 1997 Dec;38(6):1029-34.

4. Glover GH, Li TQ, Ress D. Image-based method for retrospective correction of physiological motion effects in fMRI: RETROICOR. Magn Reson Med. 2000 Jul;44(1):162-7.

5. Tijssen RH, Jenkinson M, Brooks JC, Jezzard P, Miller KL. Optimizing RetroICor and RetroKCor corrections for multi-shot 3D FMRI acquisitions. Neuroimage. 2014 Jan 1;84:394-405

6. Jones TB, Bandettini PA, Birn RM. Integration of motion correction and physiological noise regression in fMRI. Neuroimage. 2008 Aug 15;42(2):582-90

7. Cox RW. AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. 1996 Jun;29(3):162-73.

Figures

Amount of improvement in tSNR average by k-space physiological noise correction (ΔtSNR) was plotted against the total variation of IS rotation. By applying motion correction before physiological noise correction in k-space, no correlation was found between ΔtSNR and the total variation (red), whereas the data without k-space motion correction showed significant correlation between them (blue).

Cold color map indicates the voxels whose tSNR decreased after physiological noise correction. The plots show the number of those voxels of the data without k-space motion correction (left) and the data with it (right). The bar graphs show the average number of those voxels of all subjects and runs.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
3947