Sensitivity of BOLD functional MRI low frequency fluctuations to age-related brain changes
Poster No:
1140
Submission Type:
Abstract Submission
Authors:
Jessica Korte1, Wilsaan Joiner1, Audrey Fan1
Institutions:
1University of California, Davis, Davis, CA
First Author:
Co-Author(s):
Introduction:
As the global population ages, it is vital to understand healthy aging in the brain. However, within functional MRI (fMRI) analysis, there is no established metric of healthy aging. Here, we quantify reproducibility and directly compare Amplitude of Low Frequency Fluctuations (ALFF) and fractional-ALFF (fALFF) in the cerebellum, a crucial structure for learning, motor control, and coordination, on a cohort selected from Human Connectome Project (HCP) datasets [1]. ALFF and fALFF characterize the intensity of local, spontaneous BOLD fMRI fluctuations between 0.01 and 0.08 Hz, a frequency range reflective of neuronal activity [2]. Moreover, ALFF frequency sub-bands, specifically slow-3 frequencies (slow-3f) [0.073-0.198 Hz] have demonstrated similar spectral patterns to neuronal firing [3]. Slow-3f appear stronger in cerebellar areas [5], making this sub-frequency an ideal candidate to study age-related functional changes within the cerebellum.
Methods:
Data from 12 young subjects (YS; 22-35 years; 6 male) and 12 senior subjects (SS; 65-79 years; 6 male) were accessed via the HCP Young Adult S1200 and Aging datasets, respectively. Subjects successfully completed both resting-state (rs) scans without reported quality issues. MRI scans included 2 initial rs-scans and 2 follow-up rs scans. For YS, the fMRI parameters included the following: TR: 702ms, spatial resolution: 2mm isotropic, duration: 14.4 minutes, multi-band factor: 8. For SS, the parameters were: TR: 800ms, spatial resolution: 2mm isotropic, duration: 14 minutes, multi-band factor: 8.
CONN was used for preprocessing of the T1 and rs-fMRI images, and included functional realignment, slice-timing correction, normalization to MNI space and resampling of the T1 image to rs-fMRI resolution. The first 5 timepoints were discarded to account for relaxation effects. Denoising included linear regression of movement, CSF, and white matter [4]. The denoised signal was then Fourier transformed, filtered to standard (f)ALFF frequencies, and later to [0.073 0.198] Hz for slow-3f (f)ALFF. ALFF is a voxel-wise measurement, calculated by summing the amplitude across the frequency-filtered spectrum, then taking the square root of this metric. fALFF is calculated similarly but is divided by the sum of amplitudes across the full, non-filtered spectrum. Both metrics were normalized to whole-brain (f)ALFF to enable comparison across participants. All statistical analysis included cluster and multiple comparison correction and was conducted in SPM12.
CONN was used for preprocessing of the T1 and rs-fMRI images, and included functional realignment, slice-timing correction, normalization to MNI space and resampling of the T1 image to rs-fMRI resolution. The first 5 timepoints were discarded to account for relaxation effects. Denoising included linear regression of movement, CSF, and white matter [4]. The denoised signal was then Fourier transformed, filtered to standard (f)ALFF frequencies, and later to [0.073 0.198] Hz for slow-3f (f)ALFF. ALFF is a voxel-wise measurement, calculated by summing the amplitude across the frequency-filtered spectrum, then taking the square root of this metric. fALFF is calculated similarly but is divided by the sum of amplitudes across the full, non-filtered spectrum. Both metrics were normalized to whole-brain (f)ALFF to enable comparison across participants. All statistical analysis included cluster and multiple comparison correction and was conducted in SPM12.
Results:
Reproducibility analysis via paired t-tests showed high stability for (f)ALFF across rs-scans in both groups. YS demonstrated higher ALFF than fALFF in subcortical and ventricular regions, as has been shown previously [5]. SS did not demonstrate any ALFF-fALFF differences. As a result, ALFF was selected as the metric for comparison given previous demonstration of repeatability and sensitivity [5].
Slow-3f ALFF demonstrated high values (p-uncorrected<0.001, p-FWE corrected<0.05, cluster size>20) within ventricle and white matter regions relative to average whole-brain slow-3f ALFF for both groups (Fig.1), indicating sensitivity to physiological effects. Comparison between YS and SS maps of slow-3f ALFF yielded differences between senior and younger participants (p-uncorrected<0.001, p-FWE corrected<0.05, cluster size>20), as illustrated in Fig. 2.
Slow-3f ALFF demonstrated high values (p-uncorrected<0.001, p-FWE corrected<0.05, cluster size>20) within ventricle and white matter regions relative to average whole-brain slow-3f ALFF for both groups (Fig.1), indicating sensitivity to physiological effects. Comparison between YS and SS maps of slow-3f ALFF yielded differences between senior and younger participants (p-uncorrected<0.001, p-FWE corrected<0.05, cluster size>20), as illustrated in Fig. 2.
Conclusions:
This preliminary study suggests that ALFF may be a more sensitive metric than fALFF in detecting healthy aging changes. Decreased vascular reactivity and pulsatile CSF movement with age may result in smaller slow-3f ALFF in ventricles and vascular areas. Slow-3f ALFF varies with age within the cerebellum, sensorimotor areas, and frontal regions. Future work will incorporate more participants into this study, apply the SUIT atlas, and will investigate the sensitivity of other sub-frequency bands changes to aging.
Lifespan Development:
Aging 1
Novel Imaging Acquisition Methods:
BOLD fMRI 2
Keywords:
Aging
Cerebellum
FUNCTIONAL MRI
MRI
1|2Indicates the priority used for review
Provide references using author date format
[3] J. Zhang et al., “The neural correlates of amplitude of low-frequency fluctuation: a multimodal resting-state MEG and fMRI–EEG study,” Cerebral Cortex, vol. 33, no. 4, pp. 1119–1129, Feb. 2023, doi: 10.1093/cercor/bhac124.
[4] M. Woletz et al., “Beware detrending: Optimal preprocessing pipeline for low-frequency fluctuation analysis,” Human Brain Mapping, vol. 40, no. 5, pp. 1571–1582, 2019, doi: 10.1002/hbm.24468.
[5] X.-N. Zuo et al., “The oscillating brain: Complex and reliable,” NeuroImage, vol. 49, no. 2, pp. 1432–1445, Jan. 2010, doi: 10.1016/j.neuroimage.2009.09.037.
[2] Z. Yu-Feng et al., “Altered baseline brain activity in children with ADHD revealed by resting-state functional MRI,” Brain and Development, vol. 29, no. 2, pp. 83–91, Mar. 2007, doi: 10.1016/j.braindev.2006.07.002.