Multimodal neuroimaging of fatigability development

Abstract

Fatigability refers to the inability of the neuromuscular system to generate enough force to produce movements to meet task challenges. Fatigability has a central and a peripheral component linked via the neuromuscular system, but how these two components interact as fatigue develops lacks a complete understanding. The effects of fatigability are experienced in healthy humans but also accompany various disorders, often exacerbating their symptoms. We studied how fatigability develops in the neuromuscular system using multimodal neuroimaging. We recruited healthy participants to perform a fatiguing grip force task, while recording force, electromyography of forearm muscles (EMG), electroencephalography (EEG), and functional magnetic resonance imaging (fMRI) in 30-second blocks of grip task alternating with 30 seconds of rest. The task entailed maintaining 50% of the maximum force. We combined EMG and EEG to compute corticomuscular coherence and combined EEG and fMRI to compute EEG-informed fMRI. We selected eight task blocks specific to each participant to represent how the neuromuscular system adapted from pre-fatigability to actual fatigability. Those included five blocks for pre-fatigability in which participants could generate enough force to match the required 50% of maximum force and three blocks when the force fell below that limit. Across blocks of the grip force task, we observed changes in the neuromuscular system that preceded grip force changes. We found that electromyography of arm muscles shifted from high to low frequency, EEG in the channel covering the contralateral sensorimotor area increased steadily up to the fifth block and then plateaued, and fMRI signal also increased in the cerebellum. Corticomuscular coherence increased within each of the 30-second blocks of the grip task. EEG-informed fMRI revealed areas of the brain that the traditional regression did not, including the bilateral sensorimotor cortex, temporal-parietal junction, and supplementary motor area. Thus, as fatigability developed, the neuromuscular system experienced changes earlier than the actual behavior. While we found evidence for fatigability of central and peripheral origins, peripheral fatigue seems to occur first.

Keywords: EEG; EMG; corticomuscular coherence; fMRI; fatigability.

Fig. 1.

(A) Task schematic. The grip force task required participants to generate force at 50% of their maximum voluntary contraction (MVC) during 30-second blocks. They used a dynamometer with their dominant right hand. The green portion indicated the participant’s force level, while the static red horizontal bar indicated the 50% MVC. Blocks of 30-second grip were alternating with blocks of 30 seconds of rest (middle panel). A 2 seconds ‘Get Ready!’ signal appeared at the end of each rest block. (B) Hemodynamic response function (HRF). HRF for the traditional box car fMRI analysis with the eight fatigability blocks and the remaining blocks (labeled Others). (C) HRF modulated by the EEG power for the EEG-informed fMRI analysis for the eight fatigability blocks and the remaining blocks (labeled Others). First: first block; Bi, Bii, and Biii: intermediate blocks; Bn: last block before fatigability; F_n+1, F_n+2, and F_n+3, fatigability blocks. (D) Visual Analogue Scale (VAS) for the subjective evaluation of mental and physical fatigue before and after the grip force task. (E) EEG channels. Each circle represents one of the 31 channels recorded. The trapezoid indicates the EEG channels of interest.

Fig. 2.

Grip force and Visual Analogue Scale fatigue evaluation. (A) Grip force normalized to MVC of the eight blocks that represent fatigability development. The thick black line and shaded area indicate the group average ± s.e.m., respectively. Grip force remained leveled with the 50% of MVC up to the Bn block and then was significantly lower for the last three blocks. The three dots in between the first five blocks indicated discontinuity of those blocks. The first five blocks (1st to Bn) represented before-fatigability state, while the last three represented fatigability state (F_n+1, F_n+2, and F_n+3). (B) Subjective evaluation of mental and physical fatigue before and after the grip force. Physical fatigue perception, but not mental fatigue perception, was significantly higher after than before the task.

Fig. 3.

EMG analyzed with the Dimitrov Index (DI) across the eight blocks. (A) DI (means ± s.e.m.) within and across all eight blocks. DI increased steadily within and across every block. (B) DI rate of change within each block (means ± s.e.m.). Rate of change for each of the eight blocks was significantly different than zero as indicated by the */**/***. (C) DI averages of each of the eight blocks. DI increased steadily and significantly across the eight blocks. *p < 0.05, **p < 0.01, ***p < 0.001, ns: not significant.

Fig. 4.

EEG power in C3 channel across the eight blocks. (A) Spectrogram using Matlab pspectrum. The dark blue vertical bars represent the first 5 seconds of each block we omitted due to non-stationary. (B) EEG power (means ± sem) in the theta, mu and beta frequency bands. Power increased up to the Bn block and then plateaued with a slight decrease; the increase was also significant up to the F_n+3 block. (C) Rate of change of EEG power (means ± sem) within each block. EEG power increased within block for all blocks (positive slopes). F1, F2 and, F3 are the F_n+1, F_n+2, and F_n+3 blocks, respectively.

Fig. 5.

fMRI across the eight blocks. (A) Whole-brain analysis with one-way ANOVA revealed a large cluster in the posterior brain. Color represents the F statistic of the ANOVA. (B, C) BOLD signal. The large cluster in A was broken in four separate regions: precuneus, lingual gyrus, and bilateral cerebellum lobule VI. BOLD signal (means ± s.e.m.) increased significantly for all four regions. (D–H) ROI analysis. (D) ROI included the bilateral primary motor cortex (M1), pre-motor, supplementary motor area (SMA), and somatosensory cortex. (E–H) BOLD signal across the eight blocks. L: left, R: right, g: gyrus. F1, F2 and, F3 are the F_n+1, F_n+2, and F_n+3 blocks, respectively.

Fig. 6.

Corticomuscular coherence (CMC) in the beta band (13–30 Hz) between C3 and forearm muscles across the eight blocks. (A) CMC (means ± s.e.m.) within and across all eight blocks. CMC increased steadily within every block but not across blocks. (B) CMC rate of change within each block (means ± s.e.m.) always increased and was significant for seven out eight blocks, as indicated by the */**/***. (C) CMC averages of each of the eight blocks did not change significantly across the eight blocks. (D) In the beta band, higher CMC was associated with later occurrence of the Bn block, that is, the last block before fatigability. (E) In the gamma band, higher CMC was not associated with the occurrence of the Bn block. *p < 0.05, **p < 0.01, ***p < 0.001, ns: not significant.

Fig. 7.

CMC Directionality component in the efferent (brain to muscles) and afferent direction (muscles to brain) in the beta and gamma bands. (A–C) In the beta band, the average efferent direction component of each block increased across blocks. (D–F) In the beta band, the average afferent direction component increased up to the B_n block and then decreased significantly. (G–I) In the gamma band, the efferent direction component did not change significantly within or across blocks. (J–L) In the gamma band, the average afferent direction component increased from the 1^st to the F_n+1 block. ns: not significant.

Fig. 7.

CMC Directionality component in the efferent (brain to muscles) and afferent direction (muscles to brain) in the beta and gamma bands. (A–C) In the beta band, the average efferent direction component of each block increased across blocks. (D–F) In the beta band, the average afferent direction component increased up to the B_n block and then decreased significantly. (G–I) In the gamma band, the efferent direction component did not change significantly within or across blocks. (J–L) In the gamma band, the average afferent direction component increased from the 1^st to the F_n+1 block. ns: not significant.

Fig. 8.

EEG-informed fMRI between the C3 channel and BOLD signal for the eight blocks in the theta and beta bands. Color represents the group average of the EEG-fMRI association, that is, beta weights of the regression. (A) Theta band. A cluster was in the posterior brain straddling the precuneus, cuneus, and lingual gyrus, another in the right cerebellum (lobule I–IV, V, and VI and right Crus II), and another in the left cerebellum (lobule VIIIa and Crus II). (B) Regression coefficients for the theta clusters. Each dot represents one participant and the mean ± s.e.m. is shown. (C) Beta band. Clusters in the left sensorimotor areas (SM) straddling the primary motor cortex (M1) and somatosensory cortex (SS), bilateral anterior portion of the temporal-parietal junction (TPJ), supramarginal gyrus (SMG), SMA, and mid-cingulate gyrus. The dashed horizontal white line is set at z50 like the axial slice. The central and post-central sulcus were carefully drawn by hand but are approximate. (D) Regression coefficients for the beta clusters; for details see (C). L: left, R: right, g: gyrus., M1: primary motor cortex; SS: somatosensory cortex, CB: cerebellum, SMG: supramarginal gyrus, SMA: supplementary motor areas. TPJ: temporal-parietal junction.

Fig. 9.

Timeline of changes during fatigability development. (A) As grip force remained fairly constant up to the Bn block, changes in DI, RMS, EEG, and CMC efferent began much earlier and occurred together up to the Bn block. Changes in BOLD and CMC afferent occurred later. (B) Each metric (group mean ± s.e.m.) is fitted with linear regression for block 1st to Bn demonstrating that the increase was significant, but not for CMC aff. DI: Dimitrov Index, RMS: root mean square, CMC: corticomuscular coherence, eff: efferent, aff: afferent.