4236

Test-retest reliability of cerebral blood flow for assessing brain activity in a single day
Bowen Guo1,2, Tianxin Mao1,2, and Hengyi Rao1,2
1Center for Magnetic Resonance Imaging Research & Key Laboratory of Brain-Machine Intelligence for Information Behavior (Ministry of Education and Shanghai), Shanghai, China, 2School of Business and Management, Shanghai International Studies University, Shanghai, China

Synopsis

Keywords: Task/Intervention Based fMRI, fMRI

Motivation: Test-retest reliability of human activity in a single day has not yet been elucidated.

Goal(s): To determine the reliability of ASL perfusion MRI within a single day and determine the diurnal impact on the test-retest reliability of fMRI.

Approach: Participants were scanned at six time points in a day, during resting-state and Psychomotor Vigilance Test (PVT).

Results: The study revealed a higher test-retest reliability for task-based CBF compared to resting-state, with a notable diurnal decline in reliability at 20:00, and showed a decline with time intervals.

Impact: These findings underscore the importance of accounting for diurnal variations when designing functional MRI studies to ensure reliable and reproducible results, especially suggest avoiding nighttime data collection.

Introduction

Arterial spin-labeled (ASL) perfusion magnetic resonance imaging (MRI) is increasingly utilized for the evaluation of brain activity and cerebrovascular function, providing a quantitative measure of regional brain activity through absolute cerebral blood flow (CBF) values [1,2]. Although prior studies have explored the test-retest reliability of CBF across different days in resting and task states [3], however, prior research has not explored the within-day reproducibility at different time points.According to the two-process model of sleep [4,5], human performance and brain activity are modulated by circadian rhythmicity and homeostatic sleep pressure [6], and a number of studies have found the diurnal variations across diverse brain regions [7-10].
This study aims to investigate the test-retest reliability of ASL perfusion MRI within a single day while considering the potential effects of circadian rhythms and homeostatic sleep pressure.

Method

The study enrolled 38 right-handed participants with an average age of approximately 22 years, including 24 females, all with regular sleep patterns. They underwent six ASL perfusion MRI sessions at fixed time points (scheduled at 8:00 a.m., 10:00 a.m., 13:00 p.m., 15:00 p.m., 18:00 p.m., and 20:00 p.m.) throughout a single day.
All MRI scans were conducted on a 3 Tesla Siemens MAGNETOM Prisma system using a 64-channel phased-array head coil. ASL perfusion images were acquired by using a pulsed ASL (PASL) sequence with a 3D flow-sensitive alternating inversion recovery (FAIR) technique with the following parameters: repetition time (TR) = 4730 ms, echo time (TE) =16.18 ms, voxel size = 1.5×1.5×3.0 mm3, FOV= 192 × 192 mm2, matrix size= 63 × 63, slice thickness = 3 mm, inter-slice gap = 3 mm, inversion time =1990 ms, bolus duration = 700 ms.
Each session comprised both resting-state and task-based scans, with a psychomotor vigilance task designed to challenge and measure the subjects' reaction times and attention as a proxy for cerebral blood flow during task engagement. The processing process were conducted using the Statistical Parametric Mapping software (SPM 12), ASLtbx[11], DPABI V7.0 toolbox[12] and custom Matlab codes, implemented in MATLAB 2021a.

Results

For overall test-retest reliability of absolute CBF measurements throughout the day, the probability density plots of the voxel-wise ICC shared a similar characteristic of negative skew. Wilcoxon signed-rank tests showed higher ICC values for PVT scans compared to REST scans in gray matter, white matter, and all ROIs combined (all p-values < 0.001). Independent samples t-tests showed the ICC values for gray matter were significantly lower than those for white matter during PVT scans (t = -43.353, p < 0.001, Cohen’s d = 0.592) and REST scans (t = -34.395, p < 0.001, Cohen’s d = 0.470), suggesting that the test-retest reliability for white matter remained more stable throughout the day when compared to gray matter. Regarding brain networks, additionally, ICC values were consistently higher in PVT scans compared to REST scans. Contrary to absolute CBF for PVT and REST, task-induced CBF changes exhibited poor reliability in most networks.
For test-retest reliability of absolute CBF measurements at each time-point, the ICC values at 20:00 were notably lower than those at other time points in gray matter, and the majority of task-related regions and networks. However, the test-retest reliability of task-induced CBF changes at each time point remained poor for most ROIs and networks. Furthermore, regarding test-retest reliability with different interval time, as the interval between scans increased, the reliability of absolute CBF measurements tended to decrease.

Discussion

This study marks the first quantitative investigation into test-retest reliability within a single day. Our findings demonstrate that absolute CBF measures during a simple vigilance task exhibit higher reliability than that of resting-state, and the reliability of task-induced CBF changes is notably poor, in line with previous research conducted across different days [13,14].
This study provides a novel insight into the test-retest reliability of CBF measurements using ASL MRI within a single day. We found that the reliability of CBF during task performance is higher than during rest, aligning with the literature on test-retest reliability across days. Notably, gray matter showed lower reliability than white matter throughout the day. When it comes to average reliability at each time point, for absolute CBF measures, we observed that reliability dropped distinctly at 20:00 in most task-related regions and networks, and the test-retest reliability between scans during resting-state and task-state tended to decrease with the interval time.
The study highlights the influence of circadian rhythms and sleep pressure on MRI reproducibility and suggests avoiding nighttime scans for more reliable data.

Acknowledgements

No acknowledgement found.

References

[1] Buxton, R.B., Uludaǧ, K., Dubowitz, D.J., Liu, T.T., 2004. Modeling the hemodynamic response to brain activation. In: NeuroImage.

[2] Detre, J.A., Wang, J., 2002. Technical aspects and utility of fMRI using BOLD and ASL. Clin. Neurophysiol. 113, 621–634.

[3] Yang, F. N., Xu, S., Spaeth, A., Galli, O., Zhao, K., Fang, Z., ... & Rao, H. (2019). Test-retest reliability of cerebral blood flow for assessing brain function at rest and during a vigilance task. Neuroimage, 193, 157-166.

[4] Borbély, A. A. (1982). A two-process model of sleep regulation. Human Neurobiology, 1(3), 195-204.

[5] Borbély, A. A., Daan, S., Wirz-Justice, A., & Deboer, T. (2016). The two-process model of sleep regulation, a reappraisal. Journal of Sleep Research, 25(2), 131-143..

[6] Muto, V., Jaspar, M., Meyer, C., Kussé, C., Chellappa, S. L., Degueldre, C., Balteau, E., Shaffii-Le Bourdiec, A., Luxen, A., Middleton, B., Archer, S. N., Phillips, C., Collette, F., Vandewalle, G., Dijk, D.-J., & Maquet, P. (2016). Local modulation of human brain responses by circadian rhythmicity and sleep debt. Science, 353(6300), 687–690.

[7] Shannon, B. J., Dosenbach, R. A., Su, Y., Vlassenko, A. G., Larson-Prior, L. J., Nolan, T. S., Snyder, A. Z., & Raichle, M. E. (2013). Morning-evening variation in human brain metabolism and memory circuits. Journal of Neurophysiology, 109(5), 1444–1456.

[8] Xing, H., Wu, Z., Chang, Y., Ma, M., Song, Z., Liu, Y., & Dai, H. (2023). Resting‐State fMRI Study of Vigilance Under Circadian and Homeostatic Modulation Based on Fractional Amplitude of Low‐Frequency Fluctuation and Regional Homogeneity in Humans Under Normal Entrained Conditions. Journal of Magnetic Resonance Imaging, jmri.28750.

[9] Cordani, L., Tagliazucchi, E., Vetter, C., Hassemer, C., Roenneberg, T., Stehle, J. H., & Kell, C. A. (2018). Endogenous modulation of human visual cortex activity improves perception at twilight. Nature Communications, 9(1), 1274.

[10] Ly, J. Q. et al. Circadian regulation of human cortical excitability. Nat Commun 7 (2016).

[11] Wang, Z., Aguirre, G.K., Rao, H., Wang, J., Fernández-Seara, M.A., Childress, A.R., Detre, J.A., 2008. Empirical optimization of ASL data analysis using an ASL data processing toolbox: ASLtbx. Magn. Reson. Imaging 26, 261–269.

[12] Yan, C. G., Wang, X. D., Zuo, X. N., & Zang, Y. F. (2016). DPABI: data processing & analysis for (resting-state) brain imaging. Neuroinformatics, 14, 339-351.

[13] Zou, Q., Miao, X., Liu, D., Wang, D.J.J., Zhuo, Y., Gao, J.H., 2015. Reliability comparison of spontaneous brain activities between BOLD and CBF contrasts in eyes-open and eyes-closed resting states. Neuroimage 121, 91–105.

[14] Steketee, R.M.E., Mutsaerts, H.J.M.M., Bron, E.E., Van Osch, M.J.P., Majoie, C.B.L.M., Van Der Lugt, A., Nederveen, A.J., Smits, M., 2015. Quantitative functional Arterial Spin Labeling (fASL) MRI - sensitivity and reproducibility of regional CBF changes using pseudo-continuous ASL product sequences. PLoS One 10 e0132929.

Figures

Figure1. ICC metrics and activation map. First two rows denote ICC map of PVT and REST, respectively. Third and Fourth row illustrate ICC value of two task-induced CBF changes. Fifth row represents activation map of PVT(t-value), with threshold of peak-level FWE-corrected p < 0.005.

Figure2. The ICC distributions of task and resting-state CBF within GM (panel a), white matter (panel b) and all ROIs combined (panel c). PVT has higher ICC distributions than REST in all three conditions (p < 0.001), while WM has higher ICC distributions than GM either during PVT or REST.

Figure3. Median ICCs of CBF throughout the day. (A) Median ICC value during PVT and REST scans in GM, WM, and task-related ROIs. (B) Median ICC value during PVT and REST scans in networks. Median ICC value of PVT was significantly higher than that of REST in all networks. (C) Median ICC value of Delta CBFabsolute and Delta CBFproportional in GM, WM, all ROIs combined, and task-related ROIs. (D) Median ICC value of Delta CBFabsolute and Delta CBFproportional in networks.

Figure4. Average ICC value at each time point in relation to the others. (A) average ICC values of GM, WM, task-related ROIs during PVT scans. (B) average ICC values of GM, WM, task-related ROIs during REST scans. (C)average ICC values of networks during PVT scans. (D)average ICC values of networks during REST scans. (E) average ICC values of GM, WM, task-related ROIs for Delta CBFabsolute (F) average ICC values of GM, WM, task-related ROIs for Delta CBFproportional (G) average ICC values of networks for Delta CBFabsolute (H) average ICC values of networks for Delta CBFproportional

Figure5. Test-retest reliability of absolute CBF across varying time intervals. ICC value between each two scans was grouped according to their interval time. (A) Test-retest reliability of PVT across different interval time within GM, WM and task-related ROIs. (B) Test-retest reliability of REST across different interval time within GM, WM and task-related ROIs. (C) Test-retest reliability of PVT across different interval time within brain networks. (D) Test-retest reliability of REST across different interval time within brain networks.

Proc. Intl. Soc. Mag. Reson. Med. 32 (2024)
4236
DOI: https://doi.org/10.58530/2024/4236