Near-Infrared Light Increases Functional Connectivity with a Non-thermal Mechanism

Abstract

Although techniques for noninvasive brain stimulation are under intense investigation, an approach that has received limited attention is transcranial photobiomodulation (tPBM), the delivery of near-infrared light to the brain with a laser or light-emitting diode directed at the scalp. Here we employed functional magnetic resonance imaging to measure the blood-oxygenation-level-dependent signal in n = 20 healthy human participants while concurrently stimulating their right frontal pole with a near-infrared laser. Functional connectivity with the illuminated region increased by up to 15% during stimulation, with a quarter of all connections experiencing a significant increase. The time course of connectivity exhibited a sharp rise approximately 1 min after illumination onset. Brain-wide connectivity increases were also observed, with connections involving the stimulated hemisphere showing a significantly larger increase than those in the contralateral hemisphere. We subsequently employed magnetic resonance thermometry to measure brain temperature during tPBM (separate cohort, n = 20) and found no significant temperature differences between active and sham stimulation. Our findings suggest that near-infrared light synchronizes brain activity with a nonthermal mechanism, underscoring the promise of tPBM as a new technique for stimulating brain function.

Keywords: fMRI; functional connectivity; low-level laser therapy; neuromodulation; photobiomodulation.

© The Author(s) 2020. Published by Oxford University Press.

Figures

Figure 1

Experimental design. (A) To identify the effect of tPBM on hemodynamic activity in the human brain, we recorded the BOLD-fMRI signal from n = 20 healthy participants before, during, and after illumination. tPBM was applied with an 808-nm laser at an intensity of 318 mW/cm2 beginning 5 min after the start of the 25-min analysis window. The duration of tPBM was 10 min. (B) BOLD signals were converted into ROI time series corresponding to 151 cortical regions from the Destrieux atlas (Destrieux et al. 2010) and the illuminated region. (C) FC was computed as the Pearson correlation between the time courses of a “seed” region and connecting ROIs. We measured the difference in FC between the illumination and preillumination periods, and also between the postillumination and preillumination periods (indicated with light gray arrows). Permutation tests were conducted to test for significant changes in FC, with multiple comparisons corrected by controlling the FDR at 0.05.

Figure 2

Increased FC with the illuminated region during tPBM (echo 3). (A) Cortical surfaces display the group-averaged FC (Fisher transformed Pearson correlations) between the illuminated region and all left hemispheric ROIs, shown separately for the preillumination, illumination, and postillumination periods. During tPBM, increased connectivity with ROIs in the frontal and parietal lobes was readily apparent. The magnitude of the increase was reduced following illumination. (B) Same as A but showing FC with the right hemisphere. Increased connectivity with regions in the frontal, temporal, and parietal lobes was observed. (C) Bar graphs show Fisher-transformed correlation coefficients between the illuminated region and each cortical region in the left hemisphere. Error bars depict the SEM across n = 20 subjects. Connections that exhibited statistically significant increases during illumination are denoted with a black asterisk (permutation test, corrected for multiple comparisons by controlling the FDR at 0.05). In total, 19 of the 75 connections (25%) showed a significant increase. (D) Same as C but now for the right hemisphere. In total, 19 of the 75 connections exhibited a significant increase during illumination. ROI labels for all significantly enhanced connections are provided in Supplementary Table S2.

Figure 3

Increased dFC during illumination. (A) A 2-min sliding window was employed to measure dFC with the illuminated region (averaged across all 150 connecting ROIs and n = 20 subjects; shaded error bars depict the SEM across subjects). A sharp increase in dFC was observed shortly after illumination onset for all echo times. The dFC increase was somewhat reduced towards the latter portion of the illumination period—note that the sliding window extends into the postillumination period in the final 2 min of illumination. (B) To test for a possible effect of time on the observed dFC changes, we separately regressed dFC onto (i) time and (ii) a boxcar regressor modeling the time course of illumination. At all echo times, the illumination time course explained a significantly larger proportion of dFC variance: 17%/34%/27% compared with 7%/5%/<1% for echoes 1, 2, and 3, respectively (P < 1 × 10−13, Fisher r-to-z test comparing Pearson correlation coefficients between the regressor and the<?TeX \nobreak?><?TeX \hbox\bgroup?>dFC).

Figure 4

Brain-wide increases in FC during illumination (echo 3). Images show the correlation matrix between all pairs of 151 ROI time courses before (A), during (B), and after (C) illumination. The lower left and upper right quadrants indicate connections within the left and right hemispheres, respectively. The upper left and lower right quadrants indicate interhemispheric connectivity. Stimulation was delivered to the right frontal pole. (D) The difference between correlation matrices measured during and before illumination: a broad increase of up to 0.2 was readily observed, with a visible dampening of the increase in the lower left quadrant—left hemispheric connections were less affected. Binary image (right) indicates the connections that exhibited a significant increase (in white) during illumination (permutation test, corrected for 11 325 comparisons using the FDR at 0.05). A total of 283 significant connections were detected. (E) Same as (D) but now between the postillumination and preillumination periods. A total of 32 connections exhibited a statistically significant increase after illumination, with a majority of these located within the right hemisphere.

Figure 5

Accumulation of brain-wide FC throughout the illumination period. Images depict FC difference matrices between: (A) the first 5 min of illumination and the 5-min preillumination window, (B) the second 5 min of illumination and the 5-min preillumination window, (C) the second 5 min of illumination and the first 5 min of illumination, and (D) the first 5 min after illumination and the second 5 min of illumination. All matrices are shown for echo 3. Notice that, at many connections, FC is continuing to increase during the latter half of illumination.

Figure 6

Increased FC is more pronounced in the stimulated hemisphere. (A) The percent increase in FC, depicted separately for connections within the left hemisphere, between the left and right hemispheres, and within the right hemisphere. The percent change was averaged across all connections in the specified region. Error bars denote the SEM across n = 20 subjects. A repeated-measures ANOVA with hemisphere and echo time as factors and the acute FC increase as the dependent variables revealed main effects of both echo time (F(19) = 14.4, P = 2.9 × 10−20) and hemisphere (F(19) = 2.16, P = 0.0079). Post hoc paired t-tests showed that the hemispheric preference was found at echoes 2 (RH–LH vs. LH–LH; RH–RH vs. LH–LH) and 3 (RH–LH vs. LH–LH). (B) After illumination, main effects of echo time (F(19) = 12.55, P = 2.6 × 10−18) and hemisphere (F(19) = 3.71, P = 1.09 × 10−5) were also observed. Post hoc t-tests showed that the effect was focused at echo 2. (C) The number of significantly enhanced connections during illumination also exhibited a preference for the stimulated hemisphere. (D) This trend was sustained after illumination.

Figure 7

No evidence for brain heating with MR thermometry. A separate cohort of n = 20 subjects was recruited for a study aimed at resolving temperature changes during tPBM with the same dose as the BOLD study. (A) Group-averaged temperature in the illuminated region, shown separately for active (red) and sham (black) stimulation. Error bars depict the SEM across n = 20 subjects. We failed to detect any time points during which the temperature in the illuminated region was significantly different during active versus sham stimulation (paired Wilcoxon signed-rank test, n = 20, corrected for multiple comparisons by controlling the FDR at 0.05) (B) Temperature time courses for individual subjects. In all, 2 of the 20 subjects were excluded due to excessive recording artifacts. Despite the presence of drifts and spurious fluctuations, the temperature time series of active and sham stimulation largely overlapped.