Significant increases in EEG anterior-posterior alpha and beta 1 powers by transcranial photobiomodulation (tPBM) in healthy 2 humans with exclusion of thermal effects

11 Transcranial photobiomodulation (tPBM) of the prefrontal cortex can improve human cognition 12 and increase electroencephalogram (EEG) alpha and beta powers, but it was unclear whether 13 tPBM-induced heat would influence EEG oscillation powers. This study aimed to prove that 14 tPBM-induced increases in anterior-posterior EEG powers at alpha and beta bands would be 15 significant after removal of tPBM-associated thermal effects. We performed both sham-controlled 16 tPBM and sham-controlled thermal stimulation (thermo_stim) experiments under the eyes-closed 17 resting state with concurrent recordings of 64-channel EEG before, during, and after 8-min tPBM 18 at 1064-nm wavelength and thermo_stim with temperature from 33 to 41 °C, respectively, from 19 healthy humans (n=46 for tPBM; n=14 for thermo_stim). Sham-subtracted topographies of EEG 20 powers at five frequency bands were averaged at the group level during and post both stimulations. 21 Two-sample t-tests with FDR correction and effect size were calculated for comparing tPBM and 22 thermal effects at all five frequency bands. Right-frontal tPBM induced significant increases in 23 EEG anterior-posterior alpha and beta powers under the eyes-closed conditions, consistent with 24 the results previously reported in the eyes-open tPBM experiments. In contrast, right-frontal 25 thermal stimulation under the eyes-closed resting state resulted in opposite effects on EEG power 26 patterns with respect to those by tPBM. tPBM-induced enhancement in alpha and beta oscillations 27 occurred during the 8-min intervention after exclusion of thermal effects. The ability of tPBM to 28 synchronize alpha and beta oscillations in the anterior-posterior regions may be linked to the 29 enhancement of frontoparietal network and the improvement of human cognition.


Introduction 34
Photobiomodulation (PBM), also known as low-level laser therapy (LLLT) in clinical applications, 35 utilizes red to near-infrared (NIR) light to stimulate mitochondrial respiration in a wide range of 36 cells and tissues in the human body 1-4 . Transcranial photobiomodulation (tPBM) is a type of PBM 37 that delivers NIR light/laser to the human brain, which has shown promising outcomes in treating 38 psychiatric and neurological disorders 5 , such as depression and anxiety 6 , and traumatic brain 39 [CCO] as those seen by tPBM 19 . 50 There has been much less understanding and observation of electrophysiological responses to 51 tPBM in the human brain. Our recent results revealed that tPBM is effective in enhancing the EEG 52 powers of large-scale alpha and beta oscillations in the human brain during eyes-opened resting 53 state, measured by 64-channel scalp EEG from healthy human subjects 20,21 . Similar observations 54 on EEG responses to tPBM were reported by other groups while using different experimental 55

132
Designed as a single-blind, cross-over study, each subject took both sham and active tPBM 133 experiments within a period of 1 week, with a minimum of 3 days between the two experiments. 134 The order of the sham or tPBM experiment was randomly assigned. All subjects were inquired 135 about their experience after each experiment, including the heat sensation and potential drowsiness 136 they perceived. Furthermore, a thermal stimulation experiment was designed to explore the impact 137 of tPBM-associated heat sensations on human EEG signals. 138 For the thermo_stim experiment, a heat stimulator (Pathway model ATS, Pain and Sensory 139 Evaluation system, Medoc Advanced Medical Systems, Israel) was employed (see Fig. 2(a)) to 140 replicate and induce the thermal stimulation on the human forehead through an ATS mode probe. 141 The ATS thermode can deliver temperatures ranging from 0 °C to 55 °C with a maximum rate of 142

155
During the thermo_stim experiment, the thermode was placed at the same place same as where 156 the tPBM was on the right forehead to simulate/mimic the thermal effect induced by tPBM, as 157 marked in Fig. 2(b). The same 64-channel EEG device (as used in the tPBM experiment) was 158 employed to concurrently record electrophysiological responses to the thermal stimulations. Fig.  159 2(c) shows the experimental protocol: It included a 2-min baseline and an 8-min thermal 160 stimulation, followed by a 2-min recovery period. The temperature of the thermal stimulator 161 Recovery (2 minutes) remained at 33 °C during the 2-min baselines for both sham and active thermo_stim. For the active 162 stimulation, the temperature of the thermode increased from 33 °C to 41 °C following the tPBM-163 equivalent thermal rate 19 and was maintained at 41 °C during the remaining stimulation period. 164 Then we removed the thermode from the forehead during the 3-min recovery period. For the sham 165 experiment, we maintained the thermode's temperature at 33 °C throughout the 8-min period 166 before removal of the thermode from the forehead while EEG recording lasted the entire 12-min 167 period. In either experiment, subjects were asked to keep their eyes closed throughout the whole 168 measurement time. 169 While the sampling rate of EEG during the thermo_stim experiment was 512 Hz, we also 170 down sampled it to 256 Hz during data analysis to be consistent with that for the tPBM experiment. (3) where i represents either 1-4 min or 5-8 min, and f marks different frequency bands. 227

Quantification of Net EEG Power Enhancements Induced by tPBM 228
In theory, the measured percentage changes of sham-controlled mean EEG powers at respective 229 frequency bands, ∆ − , should result from both the net tPBM effect, ∆ − ( ), and 230 thermal effect, ∆ − ℎ _ , namely, 231 where i represents either 1-4 min or 5-8 min, and f marks different frequency bands. Accordingly, 233 we would be able to determine net tPBM-induced EEG power changes (in %) at respective 234 frequency bands, based on sham-controlled tPBM and thermo_stim EEG measurements (i.e., 235 ). 236

Statistical Analysis 237
Statistical analysis was performed to determine statistical significances of (1) sham-controlled 238 tPBM effects, (2) sham-controlled thermo_stim effects, and (3) the difference between sham-239 controlled tPBM and sham-controlled thermal effects. For the first two statistical testing items, 240 paired t-tests were performed between percentage topographies of mean EEG power changes (i.e., 241 mPower in %) under the sham and active tPBM as well as under the sham and thermo_stim, 242 respectively. Our cross-over designs for both tPBM and thermo_stim experiments justified the use 243 of paired t-tests. For the 3 rd statistical testing, two-sample t-tests were performed to identify 244 significant difference between the mPower values of sham-subtracted tPBM and sham-subtracted 245 thermal effects at each electrode site. Two-sample t-tests were chosen because of two different 246 groups of participants. Furthermore, the false discovery rate (FDR) corrections were performed 247 for comparison of topography to minimize type I errors in repeated t-tests among the 64 EEG 248 electrodes. All statistical tests were performed at α=0.05 after FDR correction. 249 Moreover, due to the unbalanced sample sizes between the tPBM and thermo_stim 250 experiments, the effect size (d) at each electrode was calculated for the comparison between sham-251 subtracted tPBM and sham-subtracted thermal stimulation. The "d" is defined as the difference 252 between two means divided by the standard or pooled standard deviations of the two groups. In 253 general, d = 0.2, 0.5, 0.8, and 1.2 are considered a small, medium, large, and very large effect size, 254 respectively. 255

Results 256
We reported results based on several statistical comparisons. First, we looked at the sham-257 controlled tPBM-induced effects on EEG powers at delta, theta, alpha, beta, and gamma bands. 258 Second, we investigated the sham-controlled thermo_stim effects on EEG powers at the same five 259 frequency bands. Last, we evaluated the differences between the sham-controlled tPBM versus 260 thermal effects for the respective frequency bands.

Sham-controlled Thermo_stim-induced Changes in Mean EEG Power and Topography 284
Following the same data presentation style as that in Fig. 3, The top row of Fig. 4 illustrates the 285 baseline-normalized, sham-subtracted topographies for EEG mPower alterations (in %) by 286 thermo_stim at all five frequency bands averaged over n=14 participants. The second row of Fig.  287 4 shows statistical t-maps (obtained with paired t-tests and FDR correction) during the first and 288 second halves of the stimulation period. This figure clearly demonstrates that mean EEG powers 289 in alpha and beta frequencies were significantly reduced, particularly during the last half period of 290 thermal stimulation relative to the sham condition. Specifically, an anterior-posterior reduction 291 T1-T4  T5-T8  T1-T4  T5-T8  T1-T4  T5-T8  T1-T4  T5-T8  T1-T4  T5-T8 took place in the alpha mPower while global decreases occurred in the beta mPower. However, 292 the delta, theta, and gamma bands did not show any significant changes in mPower. 293

Net EEG Power Enhancements Induced by tPBM Excluding Thermal Effects 300
As expressed by eq. (4), we would be able to quantify net tPBM-induced EEG mPower increases 301 at the group level by subtracting the sham-controlled thermo_stim mPower topography (top row 302 of Fig. 4) from the sham-controlled tPBM mPower topography (i.e., top row of Fig. 3). This 303 operation leads to Fig. 5(a), which illustrates the group-level difference of the baseline-normalized, 304 sham-subtracted spatial distributions of mPower between tPBM and thermal stimulations, i.e., 305 ∆ − ( ), during i=T1-T4 and T5-T8 at f = delta, theta, alpha, beta, and gamma bands. Figs. 306 5(b) and 5(c) are two statistical topographies for t-test (T-map) and effect sizes (ES) between sham-307 controlled tPBM effects and sham-controlled thermal effects. The t-tests were based on 2-sample 308 t-tests at α = 0.05 with FDR correction. 309 T1-T4  T5-T8  T1-T4  T5-T8  T1-T4  T5-T8  T1-T4  T5-T8  T1-T4  T5-T8 As shown in Figs. 3 and 4, thermal stimulation created the opposite effects on EEG mPower 310 patterns in alpha and beta bands with respect to those by tPBM. This opposite effect gave rise to a 311 larger difference between the two types of stimulations (i.e., ∆ − ( )), as illustrated by Fig.  312   5(a), mainly at f = alpha and beta bands. Statistically, the two-sample t-tests confirmed that the 313 alpha and beta mPower changes induced by tPBM were significantly higher with larger effect sizes 314 than those by thermo_stim at anterior-posterior regions during the entire 8-min stimulation period. 315 However, no significant photobiomodulation effects took place in delta, theta, and gamma bands. 316

Discussion 323
We recorded scalp EEG in vivo before, during, and after tPBM/sham from 49 human subjects 324 under the eyes-closed resting state. Furthermore, we generated the same temperature enhancement 325 induced by tPBM on human foreheads using a thermal generator. For the latter, we recorded EEG 326 in vivo from 14 human subjects before, during, and after the thermal stimulation with its 327  T4  T5-T8  T1-T4  T5-T8  T1-T4  T5-T8  T1-T4  T5-T8  T1-T4  T5-T8 corresponding sham experiment. Baseline-normalized EEG power alterations were compared 328 between tPBM laser-illumination and its sham experiments. More importantly, the sham-329 subtracted, tPBM-induced EEG mPower topographies from 46 participants were compared at the 330 group level with the sham-subtracted thermal-induced EEG mPower taken from another group of 331 14 human subjects. In this way, we rigorously investigated (1) how tPBM modulates brain rhythm 332 powers under the eyes-closed resting state, (2) whether and how thermal stimulation modulates 333 brain rhythm powers under the eyes-closed resting state, and (3) whether thermal stimulation effect 334 could be removed to recover/obtain net tPBM effects on the brain rhythm powers. 335 While the alpha power is believed to be related to wakefulness 28 , it is also commonly 342 associated with cognition-related brain functions such as memory encoding, attention, and brain 343 network synchronization and interaction 29-31 . Moreover, studies indicate that cortical alpha waves 344 are engendered due to the collaboration of thalamocortical and cortico-cortical interactions 32 . With 345 several hundred human subjects, previous studies have demonstrated that 1064-nm laser enabled 346 significant behavioral improvements in cognitive functions using the same experimental protocol 347 3,[9][10][11]33 . Putting all these results together, we speculate that improvement of human cognition by 348 tPBM may be closely associated with alpha power increases and potential stimulation to the 349 anterior-posterior network, which is an executive network that assists rapid instantiation of new 350 tasks by interacting with other control and processing networks 34  waves are a sign of somatosensory processing, beta activation might be responsible for the heat 358 sensation from the tPBM laser 37 . In our case, however, the beta response to heat would most 359

tPBM-induced Alterations in EEG mPower at Alpha, Beta, and Delta Bands 336
probably be limited to the left central cortex near the somatosensory region because the right 360 forehead was illuminated by the tPBM laser 38 . Thus, thermal sensation would not account for the 361 observation of beta activation beyond the somatosensory area (Fig. 3). There must exist another 362 mechanism of action for beta mPower increase, as discussed in Section 4.5. 363 Besides increasing alpha and beta mPower, tPBM reduced delta power during the first 4 364 minutes of stimulation. While the power of delta waves has been widely related to human sleep, 365 the power of delta oscillation at resting state has also been linked to cognitive functions.

Alterations in EEG mPower by Thermo_stim Are Opposite to Alterations by tPBM 376
As shown in Fig. 4, the thermal stimulation following the equivalent temperature rise given by 377 tPBM, induced significant decreases of global alpha and beta powers, meaning strong 378 desynchronizations of alpha and beta waves across the entire scalp. These observations are very 379 consistent with previous EEG studies using non-noxious thermal stimuli 45 and noxious thermal 380 stimuli 46,47 . However, few EEG studies have reported effects of nonpainful thermal stimuli given 381 on the human head since most of thermal stimulation sites were on peripheral locations 45,47-50 . One 382 study on tonic pain using continuous EEG to predict subjective pain perception observed 383 significant decreases in alpha (7-10 Hz) power during the stimulation and suggested that this 384 decrease was due to an augmented activity of cortico-cortical and thalamocortical feedback loops 385 50 . There are numerous EEG-based publications to investigate mechanisms of pain, but they are 386 beyond the focus of this study. The major observation and conclusion drawn from Figs. 3 and 4 387 were that the trends and topographic patterns of percentage changes in EEG mPower induced by 388 thermo_stim were opposite to those by tPBM. In other words, these results confirmed 389 unambiguously that the percentage changes in EEG mPower at alpha and beta bands by sham-390 controlled 1064-nm tPBM could not stem from the thermal impact of the laser used in tPBM. 391

Net tPBM Effects Excluding Thermal Effects 392
Because of the opposite trends in percentage changes of EEG mPower between the two types of 393 stimulations, as large as ~25% increases in alpha mPower (Fig. 5(a)) and large effect sizes of 0.8 394 or larger (Fig. 5(c)) were achieved in an anterior-posterior pattern under tPBM during the entire 8-395 min period relative to the sham. Furthermore, the two-sample t-tests (Fig. 5(b)) statistically proved 396 that the two stimulation methods by laser and heat created distinct changes of electrophysiology 397 in alpha and beta frequencies across most of the electrodes sites. This conclusion is in excellent 398 agreement with one of our previous studies, which presented that the thermal effect was to another mechanism or electrophysiological path, which will be explored and examined in our 405 future studies. 406 In addition, the laser/thermal stimulation applied in this study has been proved safe, non-407 painful, and often little perceptible to human subjects at the laser power of ~250 mW/cm 2 . A study 408 conducted on a rabbit brain using CW and pulsed lasers demonstrated that the heat generated by a 409 laser with less than 750 mW/cm 2 does not cause tissue damage 51 . 410

tPBM-induced EEG mPower Changes in Eyes-open and Eyes-closed Resting State 411
One of the objectives of this study was to compare the tPBM-induced effects between eyes-open 412 and eyes-closed resting state because EEG signals have been shown to behave differently between 413 these states 26,52 . In the eyes-open resting state, the human brain encounters many visual stimuli 414 and activates the visual information processing networks/paths. However, those processes and 415 pathways are suppressed during the eyes-closed resting state due to the blockage of visual 416 information input 52 , indicating the distinct emphasis of brain networks and processes in these two 417 states. This is why the eyes-closed state naturally creates higher absolute powers of alpha wave 418 optical power for tPBM delivered by the same laser. In the current study, to minimize the 437 possibility of subjects being drowsy during the experiment, we designed an 8-min stimulation at 438 3.5 W, with the same total optical energy as in the 11-min tPBM study. The current study utilized 439 a little higher laser power compared to the previous one, possibly leading to more heat/warm 440 sensation by the participants. As shown in Fig. 4, the cross-subject topographies of thermal 441 stimulation may reveal potential thermal sensation because of the reduced delta power at the frontal 442 region. 443

Methods of EEG Power Analysis, Post-Stimulation, and Frontoparietal Network 444
We employed the root-mean-square (RMS) method to quantify EEG powers during three separate 445 temporal segments (i.e., 2-min baseline, 1-4 min, and 5-8 min during tPBM) at each of the five 446 frequency bands. Conventionally, EEG powers are quantified by converting the time-domain data 447 to the frequency-domain power spectrum density (PSD) via Fourier transform, followed by 448 spectral average within the selected spectral range. Indeed, we have tested the results using both 449 RMS and PSD for several EEG time series under sham and active tPBM; the results confirmed 450 that the outcomes from both methods were in excellent agreement, as shown in supplementary 451 material, Fig. S1. 452 In this study, we did not present post-stimulation results in either tPBM or thermo_stim 453 experiments. Our focus was to address (1) whether there existed any thermal impact, created by 454 1064-nm tPBM equivalent, on the changes of EEG powers at five frequency bands, and (2) 455 whether the tPBM-evoked net increases in EEG powers at alpha and beta waves were unaffected 456 after removal of the thermal impact on EEG powers. Thus, it was not necessary to inspect results 457 in the post-stimulation period. 458 Overall, our results presented strong and significant enhancement by tPBM in EEG power or 459 synchronization for frontal-parietal alpha and beta oscillations. It is acknowledged 34 that the 460 frontoparietal network is a flexible hub for cognitive control and "a distinct control network, in 461 part functioning to flexibly interact with and alter other functional brain networks. This network 462 coordination likely occurs in a 4 Hz to 73 Hz θ/α rhythm, both during resting state and task state." 463 Thus, it is reasonable to speculate that the ability of tPBM to strongly modulate or synchronize 464 alpha and beta oscillations in the frontoparietal network may be closely associated with or serves 465 as the electrophysiological mechanism of action that tPBM is able to significantly improve human 466 cognition observed by our group [3, 7-9] and others 4,7,8,24 . 467 Moreover, according to 34 , "precision mapping of individual human brains has revealed that 468 the functional topography of the frontoparietal network is variable between individuals, 469 underscoring the notion that group-average studies of the frontoparietal network may be obscuring 470 important typical and atypical features." This notion explains why the observed spatial distribution 471 of enhanced EEG alpha and beta mPower was rather spread across frontal-parietal regions, in 472 addition to a systematic backwards shift of the EEG cap. 473

Limitations and Future Work 474
This study also had several drawbacks and thus have opportunities for future work. First, the 475 thermal stimulation was given based on contact delivery from the thermode to the human forehead, 476 whereas equivalent heat emitted from the laser in tPBM was non-contact. Also, the total area of 477 thermal stimulation by the thermode was relatively smaller than the site of tPBM laser aperture. 478 Second, the two sample sizes for tPBM and thermo_stim experiments were too unbalanced with 479 the thermal group having too fewer participants (n=14), which may cause inaccurate or insufficient 480 statistical conclusions. Last, the international 10-10 EEG cap system in this study was not strictly 481 followed since a clear area with 4-cm in diameter was needed for tPBM light delivery on the right 482 forehead. Thus, the 64-channel EEG cap was shifted about 1-2 cm backwards, which would create 483 ~1-2 cm location errors in standard 64-channel topographies given in Figs. 3-5. As for future work, 484 a non-contact heat generator may be a better option to replicate the findings in this study with a 485 larger number of human subjects. To obtain/mark more accurate mPower topography with correct 486 electrode locations, a 3-dimensional digitizer can be utilized to quantify exact locations or 487 coordinates of the 64 electrodes with respect to those in the 10-10 system on each subject's head, 488 followed by modification or correction of the EEG power topography using interpolation and 489 extrapolation based on the standard 10-10 electrode system. Last, more quantitative analysis on 490 network connectivity and directional information flow will be taken to substantiate our expectation 491 that tPBM indeed modulates the frontoparietal network significantly during and post tPBM. 492

Conclusion 493
This study demonstrated that baseline-normalized, sham-controlled tPBM with a 1064-nm laser 494 given on the right forehead of healthy human subjects neuromodulated delta, alpha, and beta 495 oscillations in eyes-closed resting state. Moreover, we demonstrated that thermal stimulations 496 would generate opposite percentage changes in alpha and beta oscillation powers with respect to 497 those by tPBM. After careful two-sample t-tests, we proved our hypothesis that tPBM-induced 498 increases in anterior-posterior EEG powers at alpha and beta bands remained statistically 499 significant during the eyes-closed resting state, similar to those during the eyes-open resting state, 500 after removal of tPBM-associated thermal effects. The observed strong enhancement by tPBM on 501 alpha and beta oscillations in the anterior-posterior regions may be the underlying 502 electrophysiological mechanism of action to explain why tPBM enables to improve human 503 cognition. 504 505