Introduction

Resting-state functional MRI (rsfMRI) has become a robust approach widely used to examine the resting state brain networks in preclinical and clinical studies. However, most preclinical neuroimaging studies are performed on sedated animals with anesthetics like isoflurane which is known to suppress the neuronal activation of the brain substantially. In order to improve the rsfMRI signal detection in anesthetized animals, ultra-high field MRI is applied together with optimal anesthesia procedures to minimize the neuronal suppression effects, and combined usage of medetomidine and isoflurane is suggested to be an optimal option for rsfMRI in rodent models [1]. However, medetomidine is not recommended for routine usage in monkeys due to its side effects in non-human primates (NHPs). Ultra-high field MRI scanners are usually not accessible for most neuroimaging studies with large animals. rsfMRI studies in NHPs are still hindered by the usage of popular anesthetics like isoflurane and ketamine. Previous studies have demonstrated that multiband MRI technique can improve the rsfMRI detection sensitivity and statistical power substantially in studies of human connectome [2]. In the present study, the multiband MRI technique was explored for rsfMRI examination in anesthetized macaque monkeys using a clinical setting.

Methods

Adult rhesus monkeys (7-11 years old, 8-10kg) were anesthetized with 1% isoflurane mixed with 100% oxygen. Physiological parameters such as O₂ saturation, blood pressure, heart rate, respiration rate, body temperature and PaCO₂ were monitored continuously and maintained in normal ranges [3]. rsfMRI data were acquired using the multiband EPI sequence (TR=0.590ms, 1000s, 2000s, TE=25ms, FOV = 96 mm × 96 mm, spatial resolution= 1.5×1.5×1.5mm³, 34 contiguous slices to cover the whole brain, multiband factor = 2, 4, GRAPPA factor = 2). A regular single-shot EPI sequence (TR/TE=2190ms/25ms, FOV = 96 mm × 96 mm, spatial resolution= 1.5×1.5×1.5mm³, 34 contiguous slices) was conducted for comparison purpose. The rsfMRI scan started ~15 minutes later after animals were moved into the scanner (Siemens 3T Trio with a 15-channel Tx/Rx volume coil). Corresponding 3D T1 weighted images and field map images were acquired. rsfMRI data were preprocessed for image distortion correction with FSL (http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FUGUE). Slice timing correction, rigid body registration, regressing out signal in white matter and cerebrospinal fluid time series with a general linear model, temporal filtering with 0.009 Hz ~0.0237 Hz band-pass, spatial smooth by a Gaussian blur with 2.5-mm full width at half maximum were performed using a script from AFNI (http://afni.nimh.nih.gov). Anatomical regions of interest (ROI) corresponding to the whole posterior cingulate cortex (PCC) were selected using AFNI and the monkey brain atlas. The averaged time course of rsfMRI signal in PCC was used for seed-based correlation analysis separately. Z transformation was applied to the individual correlation maps to show normalized correlation maps. The number of activation voxels in each normalized correlation maps (z map) was examined with AFNI script for statistical differences. All statistical analyses were performed in SPSS 21.0. P-values less than 0.05 were considered statistically significant.

Results

The signal to noise ratio (SNR) changes of EPI images acquired with the multiband EPI sequence with different TR were illustrated in Fig 1, showing dramatic SNR decrease in raw EPI images due to reduced TR. However, the rsfMRI detection sensitivity and statistical power were improved substantially due to the short TR with z >0.65 in anesthetized monkeys (Figure 2 and 3).

Discussion and Conclusion

In the present study, the multiband MRI technique was explored to improve the rsfMRI signal detection in anesthetized monkeys. It is found the reduced TR with multiband EPI decreased the SNR of raw EPI images as expected, but the sensitivity and statistical power in examining the default mode network (DMN) of anesthetized monkeys can be improved substantially with optimized TR and acceleration factor. The simultaneous multi-slice acquisition using multiband MRI technique can improve the blood-oxygen-level dependent (BOLD) acquisition in task based fMRI dramatically as shown previously [4]. Also the effect of multiband EPI acquisition for rsfMRI has been evaluated previously in human subjects using a clinical 3T scanner with 32 channel head coil [5]. Our study demonstrated that the multiband MRI technique has similar effects in the rsfMRI examination of anesthetized monkeys as seen in previous studies of conscious human subjects. Our preliminary results suggest multiband MRI with optimized scanning parameters can dramatically improve the rsfMRI detection sensitivity in anesthetized macaque monkeys using a regular clinical 3T setting.

Acknowledgements

The authors thank CMRR at University of Minnesota for providing the multiband EPI sequence. The project was funded by the National Center for Research Resources (P51RR000165) and is currently supported by the Office of Research Infrastructure Programs (OD P51OD011132).

References

[1]Grandjean et al, Neuroimage(2014); [2] Feinberg et al., PLoS ONE (2010); [3] Li et al., Brain Connectivity (2016); [4] Chen et al, Neuroimage(2015) [5] Preibisch et al., PLoS ONE (2015).