Pilot Study on Dose-Dependent Effects of Transcranial Photobiomodulation on Brain Electrical Oscillations: A Potential Therapeutic Target in Alzheimer’s Disease

Abstract

Background:
Transcranial photobiomodulation (tPBM) has recently emerged as a potential cognitive enhancement technique and clinical treatment for various neuropsychiatric and neurodegenerative disorders by delivering invisible near-infrared light to the scalp and increasing energy metabolism in the brain.

Objective:
We assessed whether transcranial photobiomodulation with near-infrared light modulates cerebral electrical activity through electroencephalogram (EEG) and cerebral blood flow (CBF).

Methods:
We conducted a single-blind, sham-controlled pilot study to test the effect of continuous (c-tPBM), pulse (p-tPBM), and sham (s-tPBM) transcranial photobiomodulation on EEG oscillations and CBF using diffuse correlation spectroscopy (DCS) in a sample of ten healthy subjects [6F/4 M; mean age 28.6±12.9 years]. c-tPBM near-infrared radiation (NIR) (830 nm; 54.8 mW/cm2; 65.8 J/cm2; 2.3 kJ) and p-tPBM (830 nm; 10 Hz; 54.8 mW/cm2; 33%; 21.7 J/cm2; 0.8 kJ) were delivered concurrently to the frontal areas by four LED clusters. EEG and DCS recordings were performed weekly before, during, and after each tPBM session.

Results:
c-tPBM significantly boosted gamma (t = 3.02, df = 7, p < 0.02) and beta (t = 2.91, df = 7, p < 0.03) EEG spectral powers in eyes-open recordings and gamma power (t = 3.61, df = 6, p < 0.015) in eyes-closed recordings, with a widespread increase over frontal-central scalp regions. There was no significant effect of tPBM on CBF compared to sham.

Conclusion:
Our data suggest a dose-dependent effect of tPBM with NIR on cerebral gamma and beta neuronal activity. Altogether, our findings support the neuromodulatory effect of transcranial NIR.

Full text

Uncorrected Author Proof
Journal of Alzheimer’s Disease xx (2021) x–xx
DOI 10.3233/JAD-210058
IOS Press
1
Pilot Study on Dose-Dependent Effects of
Transcranial Photobiomodulation on Brain
Electrical Oscillations: A Potential
Therapeutic Target in Alzheimer’s Disease
1
2
3
4
Vincenza Speraa,b,1, Tatiana Sitnikovaa,c,d,1, Meredith J. Wardi, Parya Farzamc,e, Jeremy Hughesc,e,
Samuel Gazeckia, Eric Buia,d, Marco Maielloa,b, Luis De Taboadaf, Michael R. Hambling,h,
Maria Angela Franceschinic,eand Paolo Cassanoa,d,∗
5
6
7
aDepartment of Psychiatry, Massachusetts General Hospital, Boston, MA, USA8
bDepartment of Clinical Experimental Medicine, Psychiatric Unit, University of Pisa, Pisa, Italy9
cHMS/MGH Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital,
Charlestown, MA, USA
10
11
dDepartment of Psychiatry, Harvard Medical School, Boston, MA, USA12
eDepartment of Radiology, Massachusetts General Hospital, Boston, MA, USA13
fLiteCure LLC, New Castle, DE, USA14
gLaser Research Centre, Faculty of Health Science, University of Johannesburg, Doornfontein, South Africa15
hRadiation Biology Research Center, Iran University of Medical Sciences, Tehran, Iran16
iSchool for Social Work, Smith College, Northampton, MA, USA17
Accepted 19 April 2021
Abstract.
18
Background: Transcranial photobiomodulation (tPBM) has recently emerged as a potential cognitive enhancement technique
and clinical treatment for various neuropsychiatric and neurodegenerative disorders by delivering invisible near-infrared light
to the scalp and increasing energy metabolism in the brain.
19
20
21
Objective: We assessed whether transcranial photobiomodulation with near-infrared light modulates cerebral electrical
activity through electroencephalogram (EEG) and cerebral blood flow (CBF).
22
23
Methods: We conducted a single-blind, sham-controlled pilot study to test the effect of continuous (c-tPBM), pulse (p-tPBM),
and sham (s-tPBM) transcranial photobiomodulation on EEG oscillations and CBF using diffuse correlation spectroscopy
(DCS) in a sample of ten healthy subjects [6F/4 M; mean age 28.6 ±12.9 years]. c-tPBM near-infrared radiation (NIR)
(830 nm; 54.8 mW/cm²; 65.8 J/cm²; 2.3 kJ) and p-tPBM (830 nm; 10 Hz; 54.8 mW/cm²; 33%; 21.7 J/cm²; 0.8 kJ) were
delivered concurrently to the frontal areas by four LED clusters. EEG and DCS recordings were performed weekly before,
during, and after each tPBM session.
24
25
26
27
28
29
Results: c-tPBM significantly boosted gamma (t= 3.02, df = 7, p< 0.02) and beta (t= 2.91, df = 7, p< 0.03) EEG spectral
powers in eyes-open recordings and gamma power (t= 3.61, df = 6, p< 0.015) in eyes-closed recordings, with a widespread
increase over frontal-central scalp regions. There was no significant effect of tPBM on CBF compared to sham.
30
31
32
1These authors contributed equally to this work.
∗Correspondence to: Paolo Cassano, MD, PhD, Department
of Psychiatry, Harvard Medical School, 149 13th Street (2612),
Boston, MA 02129, USA. Tel.: +1 617 643 9622; Fax: +1 617 726
5780; E-mail: pcassano@mgh.harvard.edu.
ISSN 1387-2877/$35.00 © 2021 – IOS Press. All rights reserved.

Uncorrected Author Proof
2V. Spera et al. / Transcranial NIR on EEG
Conclusion: Our data suggest a dose-dependent effect of tPBM with NIR on cerebral gamma and beta neuronal activity.
Altogether, our findings support the neuromodulatory effect of transcranial NIR.
33
34
Keywords: Cerebral blood flow, EEG oscillations, light-emitting diode, photobiomodulation, transcranial light therapy,
transcranial near-infrared light
35
36
INTRODUCTION33
Transcranial photobiomodulation (tPBM) with34
near-infrared radiation (NIR) is a novel intervention35
based on the use of low-level lasers or light-emitting
36
diodes (LEDs) that has recently emerged as a poten-37
tial valuable therapy for a range of neuro-psychiatric38
conditions. tPBM may have therapeutic effects in39
subjects with stroke, traumatic brain injury, neurode-40
generative disorders, and major depressive disorder41
[1], as well as pro-cognitive benefits in healthy popu-42
lations [1–3]. Recent findings have pointed out the43
beneficial effect of tPBM in the augmentation of
44
cognitive functions, such as memory and attention,45
in addition to emotional functions. In a placebo-46
controlled randomized study in healthy volunteers,
47
Barrett and Gonzalez-Lima [2] showed a significant48
improvement in performance on the frontal lobe cog-
49
nitive tasks and in the emotional state after tPBM50
(1064 nm laser, 250 mW/cm², 60 J/cm², 13.6 cm² x 251
sites) delivered with continuous light to participants’
52
forehead.53
The mechanisms through which tPBM with NIR
54
or red light, delivered to the scalp of patients, may55
influence the adjacent cortical areas of the brain are
56
poorly understood [4]. One promising hypothesis is
57
that tPBM may influence brain energy metabolism58
through promoting the mitochondrial function. The59
primary source of intracellular energy, adenosine60
triphosphate (ATP), which is critical to sustaining61
neural activity, is largely produced in mitochondria
62
through the process of oxidative phosphorylation.63
This process involves a respiratory chain of five64
enzyme complexes which, if altered, would influence
65
ATP synthesis. It has been suggested that tPBM, by66
delivering photons (energy particles) to the tissue,
67
may promote one of such complexes—cytochrome c68
oxidase or respiratory chain complex IV—restoring
69
or enhancing ATP production and leading to more70
energy available for neuronal activity [5]. Several71
studies reported an upregulation effect on cytochrome72
c oxidase (CCO) from both the LED and laser73
light therapy at or close to 830 nm, which led to
74
the neuronal increase in energy production [6, 7].
75
Mitochondria dysfunction has been suggested in 76
many common neuropsychiatric disorders [8], and 77
tPBM may enable compensation for such dysfunc- 78
tion, to restore cognitive capacity. 79
More active mitochondria would support higher 80
oxygen/glucose consumption, which might stimulate 81
cerebral blood flow (CBF) to deliver such nutrients. 82
Intriguingly, a preclinical study by Uozumi et al. 83
(2010) reported an increase of 30% of CBF after tran- 84
scranial NIR laser irradiation at 808 nm wavelength 85
for 15–45 min (at a power density of 1.6 W/cm²), 86
not related to heating [9]. In healthy human partici- 87
pants and clinical patients, cerebral oxygenation and 88
CBF were also found to increase [10–12]. Through 89
its effects on both CBF and brain metabolism [3], 90
tPBM is considered a potential non-invasive therapy 91
for cognitive impairment based on both animal and 92
human studies [13]. 93
More active mitochondria have also been demon- 94
strated to influence brain activity, especially oscil- 95
lations at high frequencies. Extensive research in 96
slice cultures of hippocampus has documented the 97
link between the mitochondrial function and fast 98
neural oscillations in the gamma band (∼30–90 Hz) 99
[14, 15]. Gamma frequency activity is believed to 100
arise from the interplay between cortical inhibitory 101
interneurons and excitatory principal neurons—the 102
mechanism being: high rates of interneuron firing 103
are required to synchronize function of principal 104
neurons [14, 16]. The high metabolic demands of 105
such interneuron activity may be linked to the activ- 106
ity of the mitochondrial respiratory chain in these 107
cells. Intriguingly, the parvalbumin-positive interneu- 108
rons, which are critical to gamma oscillations, show 109
higher levels of mitochondrial cytochrome c oxidase 110
when compared with principal neurons [17]. Thus, by 111
stimulating this mitochondrial enzyme (CCO), tPBM 112
may deliver targeted stimulation of the parvalbumin- 113
positive interneurons and would be expected to en- 114
hance gamma oscillations. 115
Gamma oscillations provide a fundamental mech- 116
anism of complex neuronal information processing 117
in the hippocampus and neocortex of mammals. 118
This type of brain electrophysiological activity has 119

Uncorrected Author Proof
V. Spera et al. / Transcranial NIR on EEG 3
been implicated in higher brain functions such as120
sensory perception [18], motor activity [19], and
121
memory formation [20], and are impaired in many122
common neuropsychiatric disorders [21]. Gamma123
oscillations represent a potential therapeutic target124
in Alzheimer’s disease (AD) as evidenced by both
125
animal model and human studies that correlated126
electroencephalography to AD specific processes. In127
animal models of AD, neuronal gamma oscillations128
are reduced before the onset of plaque formation
129
and of cognitive decline [22, 23]. Optogenetically-130
driving parvalbumin-positive interneurons at gamma131
frequencies (40 Hz) reduced levels of amyloid-␤
132
deposition [23]. Similarly, a non-invasive 40 Hz light-133
flickering regime reduced levels of amyloid-␤in
134
the visual cortex, in animal models [23]. Com-135
pared to age-matched healthy control subjects, both136
patients with mild cognitive impairment (MCI) and
137
patients with AD exhibit an increase in relative power138
of slow oscillations (electroencephalography (EEG)139
delta and theta rhythms), associated with a decrease140
in relative power of fast oscillations (EEG alpha,141
beta, and gamma rhythms) [24–26]. A reduction142
in spontaneous gamma-synchronization was found143
in patients with AD [21]. In patients with MCI,
144
global amyloid-␤deposition was inversely correlated
145
to alpha, beta, and gamma coherence, a marker of146
functional and effective connectivity [27]. Despite
147
this literature pointing to gamma oscillation as a148
potential therapeutic target for neurodegenerativedis-149
orders, only limited evidence is currently available
150
on the effects of tPBM on high-frequency neural151
oscillations in the human brain and on its ideal
152
dosimetry.
153
The goal of the current study was to elucidate154
the physiological processes that may underlie the155
pro-cognitive effects of tPBM with NIR by track-156
ing the effects of two different treatment doses on157
simultaneous recordings of neural activity oscil-158
lations and CBF. We delivered 830 nm tPBM by
159
a four LED clusters device to the forehead of160
healthy participants and employed EEG to quantify
161
electrophysiological brain oscillations and diffuse162
correlation spectroscopy (DCS) to index CBF. We163
anticipated that tPBM would stimulate the CBF and164
the electrophysiological oscillations, especially in the165
high frequency bands. We also expected that pulsed166
light (p-tPBM) would exert a stronger effect com-167
pared to continuous light therapy (c-tPBM), which
168
would be consistent with prior reports [28–30]. Sec-
169
ondary aims were to assess the safety and tolerability
170
of NIR-light therapy delivered transcranially.
MATERIALS AND METHODS 171
This single-site study, Transcranial Near-Infrared 172
Light in Healthy Subjects: a Cerebral Blood Flow 173
Study with Diffuse Correlation Spectroscopy (NIR- 174
flow), was approved by the Massachusetts General 175
Hospital (MGH) institutional review board (IRB). 176
The main sources of recruitment were email and web- 177
site ads through the Partners Health Care internal 178
portal for clinical trials. The study clinicaltrials.gov 179
identifier was NCT03740152. Study data are avail- 180
able upon request. s 181
Inclusion and exclusion criteria 182
Subjects (age 18–70) eligible for study participa- 183
tion were healthy by Structured Clinical Interview 184
for Diagnostic Statistical Manual-IV (SCID) criteria. 185
Subjects were enrolled in the study after providing 186
written informed consent. Female subjects of child- 187
bearing potential needed to consent (without any 188
element of coercion) to use a double-barrier method 189
for birth control if sexually active; pregnancy and lac- 190
tation were exclusionary. Other exclusionary criteria 191
included any current psychiatric disorder, substance 192
or alcohol use disorders (prior 6 months), lifetime 193
psychotic episodes, bipolar disorder, unstable med- 194
ical or neurological illness, recent history of stroke, 195
active suicidal and homicidal ideation. In addition, 196
the following criteria, potentially influencing light 197
penetration and safety, were also exclusionary: any 198
use of light activating drugs (prior 14 days), hav- 199
ing a forehead skin condition (such as tattoo or 200
open wound) and having a head-implant. Out of 201
eleven eligible subjects, ten subjects were followed 202
for the entire five-week study period and completed 203
a sequence of three sessions of tPBM and sham. The 204
study involved screening (week 1), 3 tPBM sessions 205
(weeks 2–4), and one follow-up visit (week 5). 206
Study design, blinding, and assessment schedule 207
The study included three sequential sessions, sep- 208
arated by at least one week, which entailed three 209
different modes of operation of the tPBM device 210
(Fig. 1). Participants received one session of contin- 211
uous light treatment (c-tPBM), followed by a session 212
with sham treatment (s-tPBM), which was followed 213
by a session with pulse light treatment (p-tPBM). 214
While the sequence (c-tPBM, s-tPBM, p-tPBM) was 215
the same for all subjects, the subjects were blind to 216
the specific mode of device operation used in each 217

Uncorrected Author Proof
4V. Spera et al. / Transcranial NIR on EEG
Fig. 1. Flowchart of the study procedures
session. Simultaneous recordings of EEG and DCS218
were conducted as subjects rested (with no cognitive219
task) before, during, and after each tPBM session.
220
EEG was also recorded while participants performed221
a working memory (2-back) task: once at baseline,222
before the c-tPBM session, and after each tPBM ses-
223
sion. NIR light was administered by a four LED224
clusters device (LiteCure®TPBM-1000) which had225
the flexibility to deliver either continuous light, pulse226
light, or sham. In both c-tPBM and p-tPBM mode227
the TPBM-1000 device delivered therapeutic NIR228
energy. In sham (s-tPBM), the device did not deliver
229
any light energy. The apparent behavior (i.e., the per-230
formance/output of all visible and audible indicators)
231
of the device in any of the three programmed treat-
232
ment modalities was identical. Because of the low233
average irradiances delivered in both c-tPBM and234
p-tPBM modes and because of heat sinks incorpo-235
rated in the device, the subjects did not experience236
any skin warming from NIR. These features of237
the design ensured that the study was single-blind.238
The device was designed to shut off automatically239
if skin warming above 41◦C was detected. Toler-240
ability was assessed through clinician inquiry and
241
through a self-report scale (SAFTEE-SI), which were
242
administered before the first session (c-tPBM), as a
243
baseline measure, and one week after each of the
244
three sessions to examine treatment-emergent side
245
effects.246
Intervention parameters (tPBM)247
NIR light at 830 nm was delivered with the TPBM-248
1000 device (Fig. 2) to four EEG sites (Fp1, Fp2, F3,
249
Fig. 2. Set up of TPBM-1000, DCS, and EEG devices, mounted
on study subject’s head∗.∗Of note that the TPBP-1000 (tPBM
device) was worn by study subjects, on top of EEG set-up and in
close proximity of the DCS sensors, only during the actual delivery
of NIR or sham intervention.
F4) targeting frontal poles and dorsolateral prefrontal 250
cortex, covering a total surface active treatment area 251
of 35.8 cm2[(11.52 cm2x2; Fp1, Fp2) + (6.38 cm2x2; 252
F3, F4)] with an irradiance of 54.8 mW/cm2(average 253

Uncorrected Author Proof
V. Spera et al. / Transcranial NIR on EEG 5
Table 1
Treatment parameters for tPBM delivered as continuous wave
(c-tPBM) and as pulse wave (p-tPBM)
c-tPBM p-tPBM
Wavelength 830 nm 830 nm
Irradiance (peak) 54.8 mW/cm254.8 mW/cm2
Pulsing rate - 10 Hz
Duty cycle 100% 33%
Average Power ∼2W ∼0.7W
Peak Power ∼2W ∼2W
Fluence 65.8 J/cm221.7 J/cm2
Duration of t-PBM session 20 min 20 min
Treatment window 35.8 cm235.8 cm2
Cumulative dose 2.3 kJ 0.8 kJ
in c-tPBM and peak in p-tPBM), an average fluence
254
of 65.8 J/cm2in c-tPBM and of 21.7 J/cm² in p-255
tPBM, and a total energy of 2.3 kJ in c-tPBM and 0.8256
kJ in p-tPBM (Table 1). The additional parameters257
used for pulsed light were the following: frequency258
10 Hz, duty cycle 33%. The duration of irradiation259
was 20 min (the 4 sites were irradiated concurrently).260
The entire sessions lasted about 1.5 h. The additional
261
time was needed to have the subject complete a urine262
toxicology test and self-report forms, to prepare the
263
subject, to place the necessary protections (e.g., gog-264
gles), to inspect the subject’s skin, to complete diffuse
265
correlation spectroscopy and EEG (before, during,266
and after tPBM), to set the tPBM devices, and to give267
the subject time to rest after the irradiation. tPBM268
was administered by licensed physicians (i.e., MDs)269
who were on study staff and trained to for the use of270
TPBM-1000.
271
Electroencephalographic activity (EEG)272
EEG is an imaging technique that can track elec-273
trophysiological activity of the brain with a high274
temporal resolution. Previously, EEG has been used275
to evaluate neurophysiological brain activity both276
during resting ‘task free’ scans, to study patterns of277
spontaneous dynamics of brain activity [31], and dur-278
ing task performance to examine how brain activity
279
is modulated by the demands of a task [32]. Fluctua-
280
tions in the EEG signal have been studied in several
281
frequency bands, including delta (0.5–3.5 Hz), theta
282
(4–8 Hz), alpha (8–12 Hz), beta (12–30 Hz), and283
gamma (>30 Hz). Such patterns of slow and fast fluc-284
tuations have been linked to different states of mental285
activity, and different sensory and cognitive processes286
[33].287
The Stat®X24 wireless EEG system (Advanced288
Brain Monitoring, Carlsbad, CA) was used for EEG289
data acquisition. The Stat X24 combines battery- 290
powered hardware with a preconfigured sensor strip 291
for recording twenty channel monopolar EEG. The 292
Stat X24 provides 19 EEG channels in accordance 293
with the International 10–20 system, including Fz, 294
F3, F4, Cz, C3, C4, P3, P4, Pz, O1, O2, T5, T3, F7, 295
Fp1, Fp2, F8, T4, and T6, plus adds POz. EEG (sam- 296
pling rate, 256 Hz, band pass filter: 0.1Hz high-pass, 297
100 Hz fifth order low-pass, referenced to linked mas- 298
toids) was recorded and analyzed during rest before 299
and after each tPBM session and during each 2-back 300
task session. For the 2-back task we were interested 301
in the average power spectral density (PSD) over the 302
entire task-performance session. Additionally, elec- 303
trocardiogram was recorded from collarbone left/ 304
right locations to enable removal of cardio-artifacts. 305
Independent component analysis (ICA) was used 306
to remove any artifact due to eye-blinks (using a 307
software package from OHBA: https://github.com/ 308
OHBA-analysis/osl-core). During each resting state 309
recording, participants reclined in a comfortable chair 310
for 5 min with eyes open (while looking at a cross 311
shown on a computer screen) and 5 min with eyes 312
closed. Participants were asked to relax and avoid any 313
eyebrow movements or clenching their jaw. Because 314
the state of brain function was expected to change 315
from before to after tPBM not only due to the light 316
treatment itself, but also due to relaxing in a darkened 317
room for a considerable time during treatment [34], 318
our primary comparison of interest was between EEG 319
PSD in post and pre tPBM recordings, after subtract- 320
ing corresponding PSD recorded on a sham treatment 321
visit. We kept the EEG cap on during the adminis- 322
tration of tPBM to ensure that EEG recordings were 323
from the same identical EEG sites on the scalp, before 324
and after applying NIR light. This spatial correspon- 325
dence enhanced our ability to detect pre/post changes 326
in the EEG signal. To prevent significant disruption 327
from tPBM to EEG recordings, we discarded the EEG 328
data during the tPBM administration (tPBM device 329
“on”). Because we hypothesized that tPBM would 330
have an effect on the brain activity in higher frequen- 331
cies, we limited our analyses to the gamma, beta, and 332
alpha bands. 333
Cerebral blood flow (DCS) 334
DCS is an optical technique that uses the tempo- 335
ral fluctuations of near-infrared light to measure CBF 336
directly and non-invasively [35–37]. DCS devices to 337
date are not under Food and Drug Administration 338
investigational device exemption (IDE) regulations, 339

Uncorrected Author Proof
6V. Spera et al. / Transcranial NIR on EEG
since they are considered non-significant risk. A340
homebuilt DCS system was used with a long-
341
coherence length 852 nm laser, 4 photon counting342
avalanche photodiode detectors, and a custom field-343
programmable gate array-correlator [38]. The DCS344
system was placed within 1-2 mm from one of the
345
tPBM sources, at either the left or right frontal pole.346
The DCS light source power at the optical probe was347
below 40 mW, and the beam diameter was>1mm348
(to comply with American National Standards Insti-
349
tute standards for skin exposure). The detectors were350
arranged at different distances from the source. In351
particular, one detector was at 5mm separation from
352
the source to detect superficial blood flow changes353
and 3 detectors were at 2.5 cm to detect blood flow
354
in deeper tissue. The DCS measurements were done355
before and after tPBM and every 1 min for 10 s dur-356
ing tPBM, by turning off the tPBM light and turning
357
on DCS source and detectors (to avoid detectors358
damage by the powerful tPBM light). Blood flow359
index at the short separation (BFi) was calculated360
for each detector by assuming a fixed absorption and361
scattering across subjects. CBFiat the three large362
separations was averaged together to reduce noise.363
Temporal changes of CBFi(large separation), of BFi
364
(short separation), and the subtraction of the two were
365
considered for the statistical analyses.366
Cognitive Task (2-back)
367
In the 2-back task, participants were presented with
368
a sequence of alphabet letters, and were asked to press
369
a button when the current letter matched the one from370
2 steps earlier in the sequence. Participants saw 125371
letters and 20 % of letters were targets. Each letter372
was presented for 1 s with 1 s ISI or interstimulus373
interval. Such 2-back task is believed to capture the374
engagement of working memory, one of the cogni-
375
tive functions supported by the dorsolateral prefrontal376
cortex. To perform this task, it is not enough to sim-
377
ply keep a representation of recently presented items
378
in short-term memory; the working memory buffer
379
also needs to be updated continuously to keep track380
of what the current stimulus must be compared to.381
In other words, the participant needs to both main-382
tain and manipulate information in working memory383
[39]. Our primary interest was in whether we could384
detect differences in the task performance after any385
of the active tPBM sessions (c-tPBM and p-tPBM),386
relative to the performance after sham (s-tPBM).387
Accordingly, participants were asked to perform 2-
388
back task after each active/sham session. To reduce389
burden on the participants, we administered the 2- 390
back task before tPBM only at the first visit (before 391
c-tPBM). We decided to present only results compar- 392
ing n-backs pre- and post-continuous tPBM, based 393
on the DCS and EEG results across the three t-PBM 394
sessions; since only continuous tPBM produced neu- 395
romodulation (see Results section). 396
To examine how well each participant performed 397
in the 2-back task, we computed both the accuracy in 398
detecting the target letters (2-back hits and the num- 399
ber of false positive responses) and reaction times. 400
Statistical analysis 401
Nonparametric Friedman’s test was performed to 402
compare the effect of different light treatments on 403
EEG with EEG channels treated as repeated mea- 404
sures. Similar to the parametric repeated measures 405
ANOVA, this test is used to detect differences in 406
treatments across multiple measurements [40, 41]. 407
Because differences were widespread across the 408
scalp, we also used parametric paired t-tests applied 409
to EEG averaged across channels for comparison. 410
Paired t-tests were also used to analyze DCS record- 411
ings and 2-back responses. Due to the pilot nature of 412
this study, since we recruited a small number of par- 413
ticipants, all statistics are reported including those at 414
0.05 level of significance (no correction for multiple 415
comparisons). Nonetheless, we note that a number 416
of tests would remain significant with Bonferroni 417
correction. 418
RESULTS 419
Table 2 describes the demographic characteristics 420
of the sample. The ten subjects (six females/four 421
males) had a mean age of 28.6 ±12.9 (SD). The 422
tPBM sessions were well tolerated without any 423
Table 2
Demographic characteristics on healthy subjects enrolled in the
study
Subject Age Gender Ethnicity Race Skin color
#1 56 F Hispanic White 5 out of 10
#2 28 F Not Hispanic White 2 out of 10
#3 49 M Not Hispanic White 1 out of 10
#4 23 F Not Hispanic Asian 4 out of 10
#5 23 F Not Hispanic White 2 out of 10
#6 24 M Not Hispanic White 1 out of 10
#7 20 F Not Hispanic Asian 2 out of 10
#8 21 F Hispanic More than 4 out of 10
one race
#9 22 M Not Hispanic Asian 2 out of 10
#10 20 M Hispanic White 1 out of 10

Uncorrected Author Proof
V. Spera et al. / Transcranial NIR on EEG 7
serious adverse event. As measured by the SAFTEE424
scale, administered one week after each session, three
425
subjects developed one or more adverse events: one426
subject (#1) experienced ‘feeling drowsy / sleepy’427
and ‘weakness / fatigue’, one subject (#5) experi-428
enced ‘trouble concentrating’ and one subject (#8)
429
reported ‘blurred vision’ and ‘nausea/vomiting’. Of430
note, our study included multimodal integration of431
several technologies and techniques, including EEG432
and DCS recording, tPBM delivery, 2-back cogni-
433
tive, and self-report assessments, which presented434
several logistical challenges. This complex design,435
while offering new, multimodal insights on the neu-
436
rophysiology of tPBM, led to the loss of data in some437
recording sessions, due to insufficient quality or to
438
participants’ time constraints. We report the number439
of included datapoints for each comparison below.440
Because our data samples for EEG are small, we
441
report results from a nonparametric Friedman’s test442
for repeated measures [40, 41]. We show scalp dis-443
tributions of observed EEG effects for illustration,444
but note that in this small sample of participants,445
differences among EEG channels did not reach sig-446
nificance. Additionally, because EEG data tend to447
be normally distributed, we also report paired t-test
448
results for comparison.
449
Electroencephalographic activity (EEG)450
Open eyes EEG recordings at rest451
We observed that continuous light treatment (c-452
tPBM), but not pulsed light treatment (p-tPBM),453
significantly influenced EEG PSD. Because state
454
changes (e.g., vigilance, drowsiness, anxiety) during455
rest influence PSD even within a 5 min long record-456
ing [42, 43], and because our participants remained
457
in resting state for over 30 min directly before post-
458
light-treatment assessment, we evaluated all resting459
state data collected on light-treatment days after sub-460
tracting values obtained on sham-treatment days to
461
control for state change dynamics during the lengthy
462
imaging/treatment sessions. We used resting state
463
data recorded before light/sham treatment on each464
day as a control for any changes in EEG PSD between465
session days [44, 45].466
Figure 3 shows the results for resting state record-467
ings with eyes open. Results for c-tPBM treatment468
(n= 8) are shown in the left panel. We computed469
pre-light-treatment (pre-tPBM) PSD by subtracting470
pre-treatment recording at the sham session from471
pre-treatment recording at the c-tPBM session. Simi-
472
larly, we computed post-light-treatment (post-tPBM)473
PSD by subtracting post-treatment recording at the 474
sham session from post-treatment recording at the 475
c-tPBM session. These PSD metrics gave us 476
the change value that was likely due to c-tPBM 477
treatment, while controlling for the sham effect of 478
merely resting in a dark room with a light-treatment 479
device applied but not active. The scalp maps on 480
top of Fig. 3 are shown to illustrate scalp topog- 481
raphy of PSD data averaged across participants for 482
alpha (8–12 Hz), beta (12–30Hz), and lower gamma 483
(30–55 Hz) EEG recordings, with pre-tPBM PSD 484
data in the bottom row and post-tPBM data in the 485
top row. A widespread increase in the activity PSD 486
is apparent particularly over frontal-central scalp 487
regions post-tPBM, relative to pre-tPBM, especially 488
in the gamma band. The spectrogram for these data, 489
averaged across all EEG channels, is shown below 490
Fig. 3, the solid lines show group average and the 491
shaded areas shows the standard error of the mean 492
across participants. The PSD increase in the post- 493
tPBM relative to pre-tPBM data appears larger for 494
higher frequencies. Statistical analyses revealed that 495
this effect reached significance for gamma (mean 496
change from ␥-PSD PRE(c-tPBM-Sham) 0.05 ±0.38 497
SD to ␥-PSD [POST(c-tPBM-Sham)] 0.29 ±0.31 SD; 498
nonparametric Friedman’s test between pre- and 499
post- with individual EEG channels as repeated 500
measurements: Chi-sq=21.65, p< 3.27e–06; paired 501
t-test with data averaged across EEG channels: 502
t= 3.02, df = 7, p< 0.02) and beta (mean change in 503
␤-PSD [PRE(c-tPBM-Sham)] 0.02 ±0.28 SD to ␤-PSD 504
POST(c-tPBM-Sham) 0.17 ±0.25 SD; Chi-sq = 15.80, 505
p< 7.03e–05;t= 2.91, df = 7, p<0.03) bands, and 506
showed some significance in the alpha band (mean 507
change from ␣-PSD PRE(c-tPBM-Sham) –0.03 ±0.30 508
SD to ␣-PSD POST(c-tPBM-Sham) 0.08 ±0.26 SD; Chi- 509
sq = 7.59, p< 0.006; t= 1.40, df = 7, p> 0.2). The 510
PSD effect was broadly distributed, and the dif- 511
ference between the EEG channels did not reach 512
significance in any frequency band (Friedman’s test, 513
p-values > 0.05). Results for pulsed-tPBM treatment 514
(n= 9) are shown in the right panel. Again, we 515
computed pre-tPBM PSD and post-tPBM PSD by 516
subtracting pre- and post-treatment recordings at the 517
sham session from pre- and post- tPBM recordings at 518
the p-tPBM session, respectively. Even though there 519
appears to be a frontal increase in PSD due to p-tPBM 520
treatment, the result did not reach significance. 521
We also examined the EEG PSD changes over 522
time while participants remained at rest during the 523
sham visit (e.g., due to changes in vigilance, anxiety, 524
drowsiness), and how such changes relate to the 525

Uncorrected Author Proof
8V. Spera et al. / Transcranial NIR on EEG
Fig. 3. Eyes-open resting state EEG (n= 8): spectrogram differences before (pre-LT) and after tPBM (post-LT) with continuous (CW) and
pulsed light (PW). Both the effects of CW and of PW on the spectrogram were estimated after subtracting the effect of sham.
light-treatment sessions. We computed a pre- versus526
post- sham PSD effect by subtracting pre-treatment527
recording at the sham session from post-treatment
528
recording at the sham session. Similarly, we com-529
puted pre- versus post- PSD effects for the c-tPBM
530
and p-tPBM sessions by subtracting pre-treatment531
recording from post-treatment recording at each532
session. We observed differences among these
533
pre/post PSD effects for gamma (nonparametric
534
Friedman’s test among sham, c-tPBM, and p-tPBM
535
treatments with EEG channels as repeated mea-
536
surements: Chi-sq = 14.28, p<0.0008) and beta537
(Chi-sq 10.82, p< 0.005) but not alpha (Chi-sq 1.68,538
p> 0.43) bands. The follow-up comparisons showed539
that EEG PSD decreased more from pre- to post-540
treatment during the sham session than during the
541
c-tPBM session for gamma (mean change in ␥-PSD
542
[POST(c-tPBM)–PRE(c-tPBM)] –0.02 ±0.32 SD and mean543
change in ␥-PSD [POST(Sham)–PRE(Sham)] –0.26 ±0.24544
SD; nonparametric Friedman’s test between c-tPBM
545
and sham sessions with EEG channels as repeated546
measurements: Chi-sq = 25.6, p< 4.20e–07 ;t= 3.60,547
df=7, p< 0.009) and beta (mean change in ␤-PSD548
[POST(c-tPBM)–PRE(c-tPBM)] –0.08 ±0.18 SD and549
mean change in ␤-PSD [POST(Sham)–PRE(Sham)] 550
–0.23 ±0.17 SD; Chi-sq = 27.00, p< 2.05e–07;551
t= 3.60, df = 7, p< 0.009) bands, with some signif- 552
icance in the alpha band (mean change in ␣-PSD 553
[POST(c-tPBM)–PRE(c-tPBM)] –0.04 ±0.13 SD and 554
mean change in ␣-PSD [POST(Sham)–PRE(Sham)] 555
–0.15 ±0.19 SD; Chi-sq=8.63, p< 0.004; t= 2.12, 556
df=7; p< 0.08) bands. This result indicates that 557
the EEG PSD changed during the 30 + min period 558
when participants remained in resting state due to 559
factors unrelated to the light treatment (e.g., change 560
in vigilance, anxiety, drowsiness), but that c-tPBM 561
counteracted this effect. No significant differences 562
were observed between EEG PSD changes during 563
the sham session and the p-tPBM session. None of 564
the differences between the EEG channels reached 565
significance in any frequency band (Friedman’s test, 566
p-values > 0.05). 567
Closed eyes EEG recordings at rest 568
Figure 4 shows the results for resting state record- 569
ings with closed eyes, which were obtained according 570
to the same procedure as with open eyes. Results 571
for c-tPBM treatment (n= 7) are shown in the left 572

Uncorrected Author Proof
V. Spera et al. / Transcranial NIR on EEG 9
Fig. 4. Eyes-closed resting state EEG (n= 7): spectrogram differences before (pre-LT) and after tPBM (post-LT) with continuous (CW) and
pulsed light (PW). Both the effects of CW and of PW on the spectrogram were estimated after subtracting the effect of sham.
panel. Again, there was a widespread increase in the573
activity power in post-tPBM relative to pre-tPBM574
scans for gamma and beta bands. In the alpha band,
575
activity power was high both pre-tPBM and post-
576
tPBM, which would be expected as alpha is known to
577
increase in power when participants close their eyes.
578
The global effect is illustrated in the lower panel of
579
the Fig. 4. Similarly to Open eyes EEG recordings at
580
rest, the effect on PSD appears progressively larger
581
with higher EEG frequencies. The effect reached sig-
582
nificance in the gamma band (mean change from583
␥-PSD PRE(c-tPBM-Sham) –0.11 ±0.31 SD to ␥-PSD584
[POST(c-tPBM-Sham)] 0.24 ±0.22 SD; nonparametric585
Friedman’s test between pre- and post- with individ-586
ual EEG channels as repeated measurements: Chi-587
sq = 31.11, p< 2.43e–08;t= 3.61, df = 6, p< 0.015),588
showed some significance in the beta band (mean
589
change in ␤-PSD [PRE(c-tPBM-Sham)] 0.01 ±0.26 SD590
to ␤-PSD POST(c-tPBM-Sham) 0.17 ±0.13 SD, Chi-591
sq=8.82, p< 0.003; t= 1.77, df = 6, p> 0.1), but was
592
not significant in the alpha band (mean change from593
␣-PSD PRE(c-tPBM-Sham) 0.20 ±0.37 SD to ␣-PSD594
POST(c-tPBM-Sham) 0.20 ±0.33 SD, Chi-sq = 0.31,595
p> 0.57; t= 0.026, df = 6, p> 0.8). Results for pulsed-596
tPBM treatment (n= 8) are shown in right panel. 597
The effect showed some significance in the gamma 598
band (mean change from ␥-PSD PRE(p-tPBM-Sham) 599
–0.10 ±0.40 SD to ␥-PSD [POST(p-tPBM-Sham)] 600
0.08 ±0.27 SD; nonparametric Friedman’s test 601
between pre- and post- with individual EEG channels 602
as repeated measurements: Chi-sq = 9.35, p< 0.003; 603
t= 1.27, df = 7, p> 0.2) and the beta band (mean 604
change in ␤-PSD [PRE(p-tPBM-Sham)] –0.08 ±0.33 SD 605
to ␤-PSD POST(p-tPBM-Sham) 0.10 ±0.16 SD, Chi- 606
sq = 4.39, p< 0.04; t= 1.5, df = 7, p> 0.1). None of 607
the differences between the EEG channels reached 608
significance in any frequency band (Friedman’s test, 609
p-values > 0.05). 610
In the analysis of differences in the over-time 611
changes across each session among the three treat- 612
ment conditions, we observed significant effects 613
between pre/post PSD effects for gamma (nonpara- 614
metric Friedman’s test among sham, c-tPBM, and 615
p-tPBM treatments with EEG channels as repeated 616
measurements: Chi-sq = 21.49, p< 2.16e–05 )but 617
not beta (Chi-sq 5.00, p< 0.09) or alpha (Chi- 618
sq 4.00, p> 0.1) bands. Additional comparisons 619
showed differences between c-tPBM and sham 620

Uncorrected Author Proof
10 V. Spera et al. / Transcranial NIR on EEG
Fig. 5. Eyes-open EEG during n-back task (n= 5): spectrogram differences before light therapy (pre-LT at rest) and after light therapy
(post-LT during n-back), with continuous (CW) and pulsed light (PW). Both the effects of CW and of PW on the spectrogram were estimated
after subtracting the effect of sham.
sessions for the gamma band (mean change in621
␥-PSD [POST(c-tPBM)–PRE(c-tPBM)] 0.09 ±0.28 SD622
and mean change in ␥-PSD [POST(Sham)–PRE(Sham)]
623
–0.26 ±0.34 SD; nonparametric Friedman’s test624
between c-tPBM and sham treatments with EEG
625
channels as repeated measurements: Chi-sq = 33.31,626
p< 7.87e–09;t= 3.27, df = 6, p< 0.02), and some sig-627
nificance in the beta band (mean change in ␤-PSD628
[POST(c-tPBM)–PRE(c-tPBM)] 0.03 ±0.22 SD and mean629
change in ␤-PSD [POST(Sham)–PRE(Sham)] –0.13 ±0.23630
SD; Chi-sq = 7.45, p< 0.007; t= 1.87, p> 0.1), but631
not in the alpha band (mean change in ␣-PSD632
[POST(c-tPBM)–PRE(c-tPBM)] 0.21 ±0.34 SD and mean633
change in ␣-PSD [POST(Sham)–PRE(Sham)] 0.21 ±0.31634
SD; Chi-sq = 0.56, p< 0.45; t= 0.03, p> 0.98). This
635
result again confirmed that the EEG PSD changed
636
during the 30 + min period when participants
637
remained in resting state due to factors unrelated to638
the light treatment, and that c-tPBM counteracted639
this effect. Some significance was also observed for640
differences in EEG PSD changes between the sham641
session and the p-tPBM session in the gamma band
642
(mean change in ␥-PSD [POST(p-tPBM)–PRE(p-tPBM)]
643
–0.03 ±0.25 SD and mean change in ␥-PSD 644
[POST(Sham)–PRE(Sham)] -0.22 ±0.39 SD; nonparamet- 645
ric Friedman’s test between c-tPBM and sham 646
treatments with EEG channels as repeated mea- 647
surements: Chi-sq = 7.47, p< 0.007; t= 1.4, p> 0.2). 648
None of the differences between the EEG channels 649
reached significance in any frequency band (Fried- 650
man’s test, p-values > 0.05). 651
EEG recordings during the cognitive task 652
Light-induced changes in EEG, recorded while 653
participants performed the 2-back task, were simi- 654
lar to the changes described for the resting scans: a 655
significant increase in power density of fast oscil- 656
lations was confirmed with c-tPBM. We used a 657
two-step approach to the analysis of EEG during 658
the 2-back task. First (step 1), we directly com- 659
pared EEG PSD obtained after each active treatment 660
session to EEG PSD obtained after the sham ses- 661
sion (always recorded during the 2-back task). In 662
Fig. 5, these differences (n= 5) are illustrated in 663
the scalp maps on the top row, showing subtrac- 664
tions of post- c-tPBM PSD minus post-sham PSD in 665

Uncorrected Author Proof
V. Spera et al. / Transcranial NIR on EEG 11
the left panel and subtractions of post-p-tPBM PSD666
minus post-sham PSD in the right panel. During 2-
667
back performance, an increase in the post-c-tPBM668
PSD relative to post-sham PSD was most evident669
over the frontal-temporal scalp sites. In the omnibus670
analysis with all three treatment types, we observed
671
differences in gamma (nonparametric Friedman’s672
test among sham, c-tPBM, and p-tPBM treatments673
with EEG channels as repeated measurements: Chi-674
sq=8.25, p< 0.02) and beta (Chi-sq = 6.63, p< 0.04)
675
bands. The overall increase in post-c-tPBM PSD, rel-676
ative to post-sham, was significant in the gamma band677
(mean change in ␥-PSD [POST(c-tPBM-Sham)] 0.1 ±0.1678
SD; nonparametric Friedman’s test between c-tPBM679
and sham sessions with EEG channels as repeated
680
measurements: Chi-sq = 13.82, p< 0.0002; t= 3.79,681
df=4, p< 0.02). There was some significance in the682
beta band (mean change ␤-PSD [POST(c-tPBM-Sham)]
683
0.1 ±0.1 SD; Chi-sq = 10.98, p< 0.0009; t= 2.42,684
df=4, p< 0.08). The effect was not significant for685
the post-p-tPBM comparison. Second (step 2), we686
examined if this difference between c-tPBM and687
sham visits was over and above any difference in the688
general state of the brain, which might have been689
session-dependent. We showed the post-c-tPBM ver-
690
sus post-sham effect (recorded during the 2-back
691
task) was larger than the pre-treatment difference692
in the PSD during the baseline resting scans with
693
eyes open between these sessions (shown in bottom-694
row scalp maps in Fig. 5). The overall increase695
(n= 4 – in the bottom left panel) was larger in the
696
post-c-tPBM versus post-sham (during the 2-back697
task) relative to the pre-c-tPBM versus pre-sham
698
rest subtraction in the gamma (mean change in ␥-
699
PSD [2-Back(c-tPBM-Sham)–PRE(c-tPBM-Sham)] 0.3 ±0.2700
SD; nonparametric Friedman’s test between post-2-701
back and pre-rest with EEG channels as repeated702
measurements: Chi-sq = 23.41, p< 1.31e–06,t= 3.20,703
df=3, p< 0.05) and beta (mean change in ␤-PSD704
[2-Back(c-tPBM-Sham)–PRE(c-tPBM-Sham)] 0.2 ±0.1 SD;705
Chi-sq = 27.6, p< 1.49e–07;t= 4.30, df = 3, p< 0.03)706
bands. This difference was not observed for the
707
pulsed-tPBM condition. Despite these results are708
obtained in a small sample of participants, there709
is consistency in the continuous-tPBM treatment710
response across the subjects, which is reflected in711
statistical significance. Even though the n-back per-712
formance of participants suggests some practice713
effect as performance slightly improves from ses-
714
sion 1 (c-tPBM) to session 2 (sham) to session 3
715
(p-tPBM), the increase in the brain activity (EEG)
716
in the post-c-tPBM relative to post-sham condition
717
Fig. 6. Performance on 2-back task (n= 6): accuracy in target
detection (%) and number of false positives (#) before and after
tPBM with continuous light.
occurred at session 1 relative to session 2; no improve- 718
ments were noticed in the post-p-tPBM (session 3) 719
relative to post sham condition (session 2). None of 720
the differences between the EEG channels reached 721
significance in any frequency band (Friedman’s test, 722
p-values > 0.05). 723
Cerebral blood flow (DCS) 724
In our participants’ sample (n= 10), there were no 725
statistically significant differences in the changes of 726
DCS signals at each session (c-tPBM, s-tPBM, p- 727
tPBM), for 2.5 cm and 5 mm separation and when 728
subtracting the short separation to discount the con- 729
tribution of superficial tissues to the measure of blood 730
flow. 731
Cognitive tests (n-Back) 732
We analyzed performance quality during the 733
2-back task. Figure 6 summarizes the levels of per- 734
formance on each 2-back task at session 1 (n=6, 735
c-tPBM). The performance was not significantly 736
changed from pre- to post- c-tPBM treatment both in 737
terms of accuracy in detecting targets (mean change 738
–4.49 ±20.84 %, t= –0.53, df = 6, p> 0.5) or reac- 739
tion times [from 711.4 ±116.2 SD to 733.9 ±61.1 740
SD (sec), t= 0.45, df = 5, p> 0.5].

Uncorrected Author Proof
12 V. Spera et al. / Transcranial NIR on EEG
DISCUSSION741
Our study presents two important findings, wor-742
thy of being discussed: 1) tPBM potentiated brain743
fast oscillations, and 2) tPBM-induced enhancement
744
of brain fast oscillation was dose-dependent. These745
findings have broad implications for the field of neu-746
romodulation with tPBM.747
tPBM potentiation of brain fast oscillations748
Delivered by a four LED clusters device in a sin-749
gle irradiation of 20 min to the forebrain, c-tPBM led750
to a significant increase in the high frequency neu-
751
ral activity, in the gamma and beta bands, believed752
to support higher-order cognition. During the rest-753
ing state scans with eyes either open or closed, this754
increase was over broad frontal-temporal regions of755
the scalp, showing scalp topography common of the756
neural activity in these bands [46]. During the 2-
757
back working memory task, the increase also reached
758
significance in the gamma and beta bands and was pri-
759
marily over the frontal sites, likely because this task760
engages the prefrontal cortex [47].761
Our results are consistent with several previous762
studies of tPBM effects on the brain function. A763
recent investigation in healthy older adults at risk of
764
cognitive decline reported an increase in the gamma765
EEG power, and a smaller increase in the beta power,
766
over bilateral temporal scalp regions, during laser767
tPBM (continuous wave, 1064nm, 250 mW/cm2,768
137.5 J/cm2, 13.6 cm2x 1 site) [48]. In the gamma769
band, this power enhancement persisted after tPBM.770
Another recent study with slightly different parame-771
ters of laser tPBM (continuous wave, 1064 nm, 9.72
772
J/cm2per minute, and 106.94 J/cm2over 11 min),
773
applied to the right forehead of healthy participants,774
showed an enhancement effect on the neural activ-775
ity in high frequency bands (alpha and beta) [49].
776
Whereas both these and the present study suggest a
777
potential neuromodulatory effect, their differences in
778
the analysis approach can explain some discrepancies779
in the findings. Our study which observed the tPBM780
effect on gamma and beta, not only compared the781
active tPBM to sham (similarly to prior studies), but782
also measured the change from pre- to post- tPBM783
(within session) thus controlling for any additional784
differences between sessions (e.g., in the quality of785
EEG setup). The study by Vargas and colleagues786
took a within-session comparison approach without
787
sham-control, which cannot partial out if participants788
were growing drowsier as they rested during tPBM. 789
The study by Wang and colleagues did compare 790
tPBM and sham EEG measurements; however, mea- 791
surements were collected on different days without 792
accounting for potential between-session differences, 793
such as in the EEG set-up, which gamma might 794
be sensitive to. Similarly, a recent randomized, 795
sham-controlled, double-blinded study conducted by 796
Zomorrodi et al. demonstrated a significant increase 797
in alpha, beta, and gamma power and a reduction 798
in delta and theta power after a single session of 799
pulsed (40 Hz) tPBM at 810 nm wavelength dur- 800
ing resting state. In this latter study both sham and 801
active mode caused a power increase in all frequency 802
bands comparing post to pre-stimulation, however the 803
active tPBM facilitated a power increase in alpha, 804
beta, and gamma oscillations compared to sham [50]. 805
Our study both replicates and adds to the findings 806
of Zomorrodi et al., as it extends the effects of 807
transcranial NIR on brain oscillations to the CW 808
mode (instead of PW at 40 Hz), to an irradiance 809
of ∼55 mW/cm2(lower than 100 mW/cm2) and to 810
wavelength of 830 nm (instead of 810 nm). 811
Ability of tPBM to influence gamma band neu- 812
ral activity in the human brain may have important 813
clinical significance. This type of brain activity has 814
been linked to performance on complex and attention- 815
demanding tasks [51], and was implicated in support 816
of diverse sensory and cognitive processes, including 817
perceptual processing, object representations, visual 818
awareness, and language [52–59]. Furthermore, it 819
was found that the presentation of a previously 820
learned stimulus evokes a stronger neural response 821
in the gamma band than that of a new stimulus, 822
suggesting a core role gamma activity may play 823
in the mechanism of memory [57]. Remarkably, 824
clinical states, such as major depressive disorder, 825
mild cognitive impairment, dementia due to AD and 826
Down syndrome may alter gamma band brain activity 827
[60–62]. 828
tPBM dose-dependent effect on fast brain 829
oscillations 830
Most of the effects of tPBM on high frequency 831
brain oscillations disappeared when pulsed light was 832
used (p-tPBM). We interpreted this change as a dose- 833
dependent effect, either due to the pulsed nature 834
of tPBM (10 Hz, 33% duty cycle), i.e., the on-off 835
cycling of the light delivered affected the response, 836
or due to the overall lower total energy delivered— 837
0.8 kJ in pulsed-tPBM compared to 2.3 KJ in 838

Uncorrected Author Proof
V. Spera et al. / Transcranial NIR on EEG 13
continuous-tPBM. A dose-dependent effect of laser839
tPBM with continuous light on behavior, accompa-
840
nied by changes in the electrocorticogram spectra,841
was previously demonstrated in an animal model842
of depression [63]. After pharmacological induction843
of depression, rats presented with both reduced sur-
844
vival behaviors on a forced swim test and reduced845
gamma-beta power at intracranial EEG. While laser846
tPBM led to normalization of behavior and electro-847
corticogram, only its lowest dose was effective; the
848
middle dose produced no behavioral changes and the849
highest worsening of depressive behaviors. Observ-850
ing a dose-dependent effect of tPBM on the neural
851
activity in humans has great scientific significance852
as this suggests a causal link. It will be important in
853
future research to further establish if different doses854
of tPBM influence the degree of the evoked neu-855
ral changes. Characterizing the dose-response curve
856
also has clinical significance, enabling the delivery of857
effective tPBM treatments. Our data suggests tPBM858
thresholds of insufficient (≤0.8 kJ) and likely effec-859
tive (≥2.3 kJ) total energy for a single session with860
low irradiance in healthy subjects.861
Lack of tPBM effect on cognition and CBF
862
Large enhancement on the resting state gamma and
863
beta power, which was consistent across participants,
864
suggests promise of tPBM for treatment of cogni-
865
tive deficits. Intriguingly, the resting state gamma866
power has previously been linked to language and
867
cognition during early development and in clinical
868
conditions such as schizophrenia. Prior studies have869
also showed the effects of a single session of tPBM on870
cognition. For instance, Barrett and Gonzalez-Lima871
(2013) found that measures of attention (psychomo-
872
tor vigilance task, PVT) and memory (delayed match
873
to sample, DMS) improved in response to tPBM. In874
our study, we employed the n-back working memory
875
task, which like PVT and DMS depends on the func-876
tion of the prefrontal cortex. Unfortunately, in this
877
pilot study the sample size was small, which is a likely878
explanation for why we failed to observe improve-879
ment on this task due to tPBM. Our very low accuracy880
rate at 2-back (67–71%), in a sample of mostly881
high-functioning and healthy subjects, also suggests882
that subjects might have underperformed, therefore883
jeopardizing the reliability of the task. Finally, it is
884
possible that tPBM with low irradiance might require
885
multiple sessions and longer follow-up to demon-
886
strate an effect on cognition.
887
It was surprising that our study found no sig- 888
nificant changes in the CBF, as indexed by DCS 889
recorded at 5 mm and at 2.5 cm source-detector sep- 890
aration in response to both active tPBM sessions, 891
compared to the sham mode. In fact, to this date, 892
one of the most validated and replicated neurophys- 893
iological findings in humans treated with tPBM is 894
the increase in CBF [11], although, prior studies, 895
demonstrating an effect of tPBM on CBF, typically 896
used devices with higher irradiance in the order of 897
75–250 mW/cm2[64–66]. Furthermore, DCS, the 898
technology used for CBF detection, is a validated 899
technique with numerous studies supporting its use in 900
humans. We believe that our measure of CBF was not 901
sensitive enough and consequently inaccurate. There 902
are several reasons for our current interpretation of 903
our DCS findings: 1) Cerebrovascular coupling is 904
expected whenever cortical brain areas are activated, 905
no matter the mechanism of neuromodulation. In 906
other terms, even if minimal photochemical reac- 907
tions occurred in cortical mitochondria, still neuronal 908
activity would consequently evoke vascular changes. 909
2) While every effort was made to place the DCS 910
detectors in close proximity to the tPBM NIR source 911
(1-2 mm), the light was distributed over four sources, 912
of which only one was chosen for adjacent placement 913
of DCS (therefore reducing the NIR dose in proxim- 914
ity of the detector). Moreover, the complexity of our 915
assembly—including four rigid NIR light sources, a 916
full EEG cap, a protective eyewear and DCS sen- 917
sors, as well as DCS fibers—is likely to have affected 918
our measure by either shielding the NIR light close 919
to the DCS sensor or by distancing the DCS sensor. 920
3) The DCS sensor measures both short (5 mm) and 921
long (2.5 cm) separation effects, which are expected 922
to reflect blood flow changes respectively within the 923
superficial tissues (skin) and within superficial and 924
deep tissues combined (including brain). Strikingly, 925
in our measures no changes in the short-separation 926
DCS signal were recorded. While we were not inter- 927
ested in the effects on skin vasculature, tPBM NIR 928
has intrinsic vasomotor effects which should have 929
resulted in a detectable DCS signal. 4) Other groups 930
have reported on tPBM effects on both brain oscil- 931
lations and CBF, when using a total energy of NIR 932
equal or less than one quarter of our study dose [65]. 933
Of note, in our study, two of the light sources (F3 934
and F4) were frequently placed above hair and this 935
might have resulted in an overestimate of the light 936
energy deposition. 5) Finally, it is unlikely that the 937
small sample size might explain the lack of tPBM 938
effect on CBF, since this effect size was negligible.

Uncorrected Author Proof
14 V. Spera et al. / Transcranial NIR on EEG
Strengths and limitations939
Several limitations should be acknowledged for940
our study: 1) This pilot study is based on a small941
sample of healthy participants; therefore, despite the
942
strong effect on brain oscillations, the current results943
can only be considered preliminary. It is unproven944
whether the effects on brain oscillation would gen-945
eralize to a wider population, including patients946
suffering brain disorders, such as AD. 2) Similarly,
947
due to the small study sample, we cannot judge con-
948
clusively the potential effects of low dose p-tPBM949
on brain oscillations; it is striking that large effects950
were observed for continuous light tPBM, suggest-
951
ing that, at the very least, effect sizes of pulsed and952
continuous light dosages were quite different. 3) In953
each session of our pilot study we recorded EEG
954
during eyes open resting state, eyes closed resting955
state and during the cognitive task with no sequence
956
randomization. 4) It is also noteworthy that we only957
used a 20-channel EEG recording; it may be nec-
958
essary to record with higher EEG sensor density to959
adequately characterize tPBM effects on brain oscil-960
lations (e.g., the effect of the p-tPBM over the anterior
961
scalp regions). 5) Because the dose of tPBM was962
much lower in pulse mode, this prevented from test-963
ing the impact of the pulsing of the light on brain964
oscillations. In the future, studies might adequately965
compare continuous and pulsed waves effects, by
966
matching average irradiance and total energy per967
session. 6) An additional limitation, also related to
968
the study device, is the use of four clusters of LED
969
sources mounted on a rigid frame. The rigid frame
970
prevented the repositioning of two sources (on F3,971
F4) below the hairline; as already mentioned, it is972
therefore likely that the actual dose of NIR was less,973
due to the shielding effect of hair in this young cohort.974
7) Our measure of the effects on cognitive func-
975
tion appears to be affected by the subjects’ poor976
effort at the n-back task, in addition to the small
977
sample size. The strengths of our approach are: 1)978
In our study, tPBM delivered with an LED device
979
was well tolerated with no serious adverse events. 2)980
LED tPBM has several advantages over laser, such981
as low or no risk of retina injury, lower cost, and982
potential self-administration. These features are deci-983
sive in terms of broadening of the clinical use of984
tPBM, as they offer a considerable progress in safety985
and comfort, compared to prior studies using laser
986
sources. 3) The multimodal neurophysiological test-
987
ing applied in our study offered a rare opportunity to
988
understand the effects of tPBM in humans. 4) The
989
single-blind, sham-controlled design, with careful 990
match of all visible and auditory outputs of the tPBM 991
device between active and sham sessions contributes 992
to rigorous testing of our hypotheses. Of note, only 993
two investigators (PC and EB) were always aware of 994
the exact mode of tPBM applied at each session (c- 995
tPBM, sham, p-tPBM), while investigators involved 996
with neurophysiological testing remained mostly 997
blind. 998
Considerations on tPBM dosimetry, penetration, 999
and mechanism of action 1000
A final consideration should be made to the postu- 1001
lated mechanism of action of tPBM, responsible for 1002
the potentiation of fast brain oscillations. While most 1003
human studies on tPBM do not assess the expected 1004
NIR penetration based on study parameters, on the 1005
sample characteristics, on the position of the light 1006
source and of the target area (through cadaver mod- 1007
els or through specific simulations), the uncertainty 1008
over photon deposition at cortical level could be con- 1009
sidered a limitation of our study. Henderson and 1010
Morries (2019) have extensively reviewed the litera- 1011
ture on penetration of the light through human living 1012
tissues and in cadaver models and they concluded 1013
that there are scenarios when the expected deposi- 1014
tion of photons to the brain is negligible [4]. They 1015
also postulated that in these cases the observed neu- 1016
rophysiological or therapeutic effects of tPBM are 1017
likely to be related to the systemic effects of NIR 1018
such as anti-inflammatory and anti-oxidant effects 1019
[4]. Our group is familiar with the therapeutic effects 1020
on brain disorders of systemic PBM [67]. As a team, 1021
we have speculated that in addition to direct effects 1022
onto the brain, associated with photon deposition 1023
to the brain, and in addition to indirect, systemic 1024
effects, likely associated with the irradiation of blood 1025
[68], there might be a third potential mechanism 1026
related to indirect but local effects. Although, we 1027
have no proof to offer, we speculated that tPBM 1028
might induce very weak skin currents which might 1029
induce changes in neurophysiology such as the ones 1030
demonstrated in our experiment, even in the absence 1031
of actual deposition of NIR onto the brain. The 1032
effects on brain oscillations of very weak skin cur- 1033
rents applied to the forehead, and their potential 1034
therapeutic effects, are well-known [69, 70]. Some of 1035
these modalities of neuromodulation are transcranial 1036
DCS, transcranial alternating current stimulation, and 1037
cranial electrotherapy stimulation [71]. While we 1038
expected in our experiment a direct effect of NIR light 1039

Uncorrected Author Proof
V. Spera et al. / Transcranial NIR on EEG 15
onto brain function, associated with photon deposi-1040
tion onto the brain, we acknowledge that this might
1041
not be the case and either an indirect systemic and/or1042
an indirect local effect might be solely responsible1043
for the observed changes in fast brain oscillations.1044
In favor of a direct effect is that our team mem-
1045
bers previously reported that a fluence of 0.3 J/cm2
1046
(810 nm) at the target tissue was indeed effective in1047
modulating neuronal metabolism (e.g., ATP produc-1048
tion) and mitochondrial function (e.g., mitochondrial
1049
membrane potential) [72]. Also, our group has shown1050
that light deposition with the current study param-1051
eters (NIR in CW mode) is expected to achieve at
1052
least 0.3 J/cm2at the cortical level [73]. We also1053
demonstrated in a follow up study that penetration
1054
of transcranial NIR light is significantly greater in1055
young adults, compared to middle-aged and older1056
adults; of note, in our current study cohort, all but two
1057
study subjects were in their twenties [74]. According1058
to these data, our CW parameters of tPBM could have1059
been barely sufficient to produce neurophysiological1060
effects based on the actual penetration of the NIR1061
light to the brain surface. However, a case could be1062
made for the opposite, in line with the work of Hen-1063
derson and Morries (2019), penetration of NIR light
1064
through scalp and skull can be as low as 1-2% of
1065
incident light and can even be none, depending on1066
selected parameters [4]. Interestingly, based on the
1067
work of Lychagov and colleagues (2006) and of Ted-1068
ford and colleagues (2015), even when using laser1069
sources and high power of NIR light there is remark-
1070
able interindividual variability in light penetration,1071
which might render direct effects onto the brain ques-
1072
tionable at times [75, 76]. Lastly, if we consider 0.9
1073
J/cm2as the threshold of photon deposition at tar-1074
get tissue, necessary to achieve a direct effect [76],1075
this would be above the average expected deposition1076
in our study; therefore, solely indirect effects should1077
be postulated for the NIR mechanism of action. One1078
additional level of complexity is that there is exten-
1079
sive variability of the expected deposition of NIR1080
light within any target area of the brain, such a Brod-
1081
mann’s area of the cortex. For instance, the upper1082
quartile of a given area (based on light deposition),1083
potentially receives 3-fold higher fluence than the1084
overall average for the same brain area [73]. Depend-1085
ing on the proportion of a given brain area needed to1086
be modulated in order to trigger macroscopic neuro-1087
physiological changes, the upper quartile fluence at
1088
target tissue rather than the average fluence could be
1089
considered to predict direct effects of NIR on brain
1090
function.
CONCLUSIONS 1091
We observed a significant and large enhancement 1092
of the power spectral density of neural activity in the 1093
gamma-band in response to c-tPBM with NIR. This 1094
result is consistent with our hypothesis that tPBM 1095
influences high-frequency synchronized brain activ- 1096
ity in the gamma band. By modulating the brain 1097
gamma activity, linked to higher-order cognition, 1098
tPBM might have a promise as a procognitive ther- 1099
apy, and more specifically as an intervention for AD. 1100
It is encouraging that a large and significant increase 1101
in EEG metrics was found at rest with eyes open 1102
and closed and during cognitive challenge, despite 1103
the small sample of participants. 1104
ACKNOWLEDGMENTS 1105
The authors acknowledge the help of Arianna Ric- 1106
cio for drawing the depiction of our Transcranial 1107
Photobiomodulation device (TPBM-1000), diffuse 1108
correlation spectroscopy, and electroencephalogram 1109
systems (Fig. 2). 1110
None of the funding sources had any involvement 1111
neither in the study design, in the collection, analysis, 1112
interpretation of study data [but in the writing of the 1113
report, Luis DeTaboada offered his comments to the 1114
final manuscript], nor in the decision to submit 1115
the article for publication. 1116
LiteCure LLC provided all funding to support this 1117
study. There is no associated grant number to provide. 1118
Authors’ disclosures available online (https:// 1119
www.j-alz.com/manuscript-disclosures/21-0058r1). 1120
REFERENCES 1121
[1] Cassano P, Petrie SR, Mischoulon D, Cusin C, Katnani 1122
H, Yeung A, De Taboada L, Archibald A, Bui E, Baer L, 1123
Chang T,Chen J, Pedrelli P, Fisher L, Farabaugh A, Hamblin 1124
MR, Alpert JE, Fava M, Iosifescu DV (2018) Transcranial 1125
photobiomodulation for the treatment of major depressive 1126
disorder. The ELATED-2 Pilot Trial. Photomed Laser Surg 1127
36, 634-646. 1128
[2] Barrett DW, Gonzalez-Lima F (2013) Transcranial infrared 1129
laser stimulation produces beneficial cognitive and emo- 1130
tional effects in humans. Neuroscience 230, 13-23. 1131
[3] Blanco NJ, Maddox WT, Gonzalez-Lima F (2017) Improv- 1132
ing executive function using transcranial infrared laser 1133
stimulation. J Neuropsychol 11, 14-25. 1134
[4] Henderson TA, Morries LD (2019) Near-infrared photonic 1135
energy penetration—principles and practice. In Photo- 1136
biomodulation in the Brain, Hamblin M, Huang Y-Y, eds. 1137
Academic Press, pp. 67-88. 1138

Uncorrected Author Proof
16 V. Spera et al. / Transcranial NIR on EEG
[5] Hennessy M, Hamblin MR (2017) Photobiomodulation and
1139
the brain: A new paradigm. J Opt 19, 013003.1140
[6] Mochizuki-Oda N, Kataoka Y, Cui Y, Yamada H, Heya M,1141
Awazu K (2002) Effects of near-infra-red laser irradiation on1142
adenosine triphosphate and adenosine diphosphate contents1143
of rat brain tissue. Neurosci Lett 323, 207-210.
1144
[7] Wong-Riley MT,Liang HL, Eells JT, Chance B, Henry MM,1145
Buchmann E, Kane M, Whelan HT (2005) Photobiomod-1146
ulation directly benefits primary neurons functionally1147
inactivated by toxins: Role of cytochrome c oxidase. J Biol
1148
Chem 280, 4761-4771.1149
[8] Marazziti D, Baroni S, Picchetti M, Landi P, Silvestri S, Vat-1150
teroni E, Catena Dell’Osso M (2012) Psychiatric disorders1151
and mitochondrial dysfunctions. Eur Rev Med Pharmacol
1152
Sci 16, 270-275.1153
[9] Uozumi Y, Nawashiro H, Sato S, Kawauchi S, Shima K,1154
Kikuchi M (2010) Targeted increase in cerebral blood flow1155
by transcranial near-infrared laser irradiation. Lasers Surg
1156
Med 42, 566-576.1157
[10] Tian F, Hase SN, Gonzalez-Lima F, Liu H (2016)
1158
Transcranial laser stimulation improves human cerebral1159
oxygenation. Lasers Surg Med 48, 343-349.1160
[11] Salgado AS, Zangaro RA, Parreira RB, Kerppers, II (2015)1161
The effects of transcranial LED therapy (TCLT) on cere-1162
bral blood flow in the elderly women. Lasers Med Sci 30,
1163
339-346.
1164
[12] Wang X, Tian F, Reddy DD, Nalawade SS, Barrett DW,
1165
Gonzalez-Lima F, Liu H (2017) Up-regulation of cerebral
1166
cytochrome-c-oxidase and hemodynamics by transcranial
1167
infrared laser stimulation: A broadband near-infrared spec-1168
troscopy study. J Cereb Blood Flow Metab 37, 3789-3802.1169
[13] de la Torre JC (2016) Cerebral perfusion enhancing inter-1170
ventions: A new strategy for the prevention of Alzheimer1171
dementia. Brain Pathol 26, 618-631.
1172
[14] Galow LV, Schneider J, Lewen A, Ta T-T, Papageorgiou1173
IE, Kann O (2014) Energy substrates that fuel fast neuronal
1174
network oscillations. Front Neurosci 8, 398-398.
1175
[15] Whittaker RG, Turnbull DM, Whittington MA, Cunning-1176
ham MO (2011) Impaired mitochondrial function abolishes1177
gamma oscillations in the hippocampus through an effect1178
on fast-spiking interneurons. Brain 134, e180-e181.
1179
[16] Buzsaki G, Wang XJ (2012) Mechanisms of gamma oscil-
1180
lations. Annu Rev Neurosci 35, 203-225.1181
[17] Gulyas AI, Buzsaki G, Freund TF, Hirase H (2006) Popu-
1182
lations of hippocampal inhibitory neurons express different
1183
levels of cytochrome c. Eur J Neurosci 23, 2581-2594.1184
[18] Gross J, Schnitzler A, Timmermann L, Ploner M (2007)1185
Gamma oscillations in human primary somatosensory cor-
1186
tex reflect pain perception. PLoS Biol 5, e133.
1187
[19] Nowak M, Zich C, Stagg CJ (2018) Motor cortical gamma
1188
oscillations: What have we learnt and where are we headed?1189
Curr Behav Neurosci Rep 5, 136-142.1190
[20] Sederberg PB, Schulze-Bonhage A, Madsen JR, Bromfield1191
EB, McCarthy DC, Brandt A, Tully MS, Kahana MJ (2007)1192
Hippocampal and neocortical gamma oscillations predict1193
memory formation in humans. Cereb Cortex 17, 1190-1196.
1194
[21] Herrmann CS, Demiralp T (2005) Human EEG gamma
1195
oscillations in neuropsychiatric disorders. Clin Neurophys-1196
iol 116, 2719-2733.1197
[22] Nakazono T, Jun H, Blurton-Jones M, Green KN,1198
Igarashi KM (2018) Gamma oscillations in the entorhinal-1199
hippocampal circuit underlying memory and dementia.
1200
Neurosci Res 129, 40-46.1201
[23] Iaccarino HF, Singer AC, Martorell AJ, Rudenko A, Gao F,1202
Gillingham TZ, Mathys H, Seo J, Kritskiy O, Abdurrob F,1203
Adaikkan C, Canter RG, Rueda R, Brown EN, Boyden ES, 1204
Tsai LH (2016) Gamma frequency entrainment attenuates 1205
amyloid load and modifies microglia. Nature 540, 230-235. 1206
[24] van der Hiele K, Vein AA, vander Welle A, van der Grond J, 1207
Westendorp RG, Bollen EL, van Buchem MA, van Dijk JG, 1208
Middelkoop HA (2007) EEG and MRI correlates of mild 1209
cognitive impairment and Alzheimer’s disease. Neurobiol 1210
Aging 28, 1322-1329. 1211
[25] Czigler B, Csikos D, Hidasi Z, Anna Gaal Z, Csibri E, 1212
Kiss E, Salacz P, Molnar M (2008) Quantitative EEG in 1213
early Alzheimer’s disease patients - power spectrum and 1214
complexity features. Int J Psychophysiol 68, 75-80. 1215
[26] Moretti DV, Pievani M, Geroldi C, Binetti G, Zanetti O, 1216
Rossini PM, Frisoni GB (2010) EEG markers discriminate 1217
among different subgroup of patients with mild cognitive 1218
impairment. Am J Alzheimers Dis Other Demen 25, 58-73. 1219
[27] Michels L, Muthuraman M, Anwar AR, Kollias S, Leh SE, 1220
Riese F, Unschuld PG, Siniatchkin M, Gietl AF, Hock C 1221
(2017) Changes of functional and directed resting-state con- 1222
nectivity are associated with neuronal oscillations, ApoE 1223
genotype and amyloid deposition in mild cognitive impair- 1224
ment. Front Aging Neurosci 9, 304. 1225
[28] Lapchak PA, De Taboada L (2010) Transcranial near 1226
infrared laser treatment (NILT) increases cortical adeno- 1227
sine-5’-triphosphate (ATP) content following embolic 1228
strokes in rabbits. Brain Res 1306, 100-105. 1229
[29] Brondon P, Stadler I, Lanzafame RJ (2009) Pulsing influ- 1230
ences photoradiation outcomes in cell culture. Lasers Surg 1231
Med 41, 222-226. 1232
[30] Lapchak PA, Salgado KF, Chao CH, Zivin JA (2007) Tran- 1233
scranial near-infrared light therapy improves motor function 1234
following embolic strokes in rabbits: An extended thera- 1235
peutic window study using continuous and pulse frequency 1236
delivery modes. Neuroscience 148, 907-914. 1237
[31] Rosazza C, Minati L (2011) Resting-state brain networks: 1238
Literature review and clinical applications. Neurol Sci 32,1239
773-785. 1240
[32] Bridwell DA, Cavanagh JF,Collins AGE, Nunez MD, Srini- 1241
vasan R, Stober S, Calhoun VD (2018) Moving beyond ERP 1242
components: A selective review of approaches to integrate 1243
EEG and behavior. Front Hum Neurosci 12, 106. 1244
[33] Herrmann CS, Struber D, Helfrich RF, Engel AK (2016) 1245
EEG oscillations: From correlation to causality. Int J Psy- 1246
chophysiol 103, 12-21. 1247
[34] Ward AM, McLaren DG, Schultz AP, Chhatwal J, Boot BP, 1248
Hedden T, Sperling RA (2013) Daytime sleepiness is asso- 1249
ciated with decreased default mode network connectivity in 1250
both young and cognitively intact elderly subjects. Sleep 36,1251
1609-1615. 1252
[35] Mesquita RC, Durduran T, Yu G, Buckley EM, Kim MN, 1253
Zhou C, Choe R, Sunar U, Yodh AG (2011) Direct mea- 1254
surement of tissue blood flow and metabolism with diffuse 1255
optics. Philos Trans A Math Phys Eng Sci 369, 4390-4406. 1256
[36] Durduran T, Yodh AG (2014) Diffuse correlation spec- 1257
troscopy for non-invasive, micro-vascular cerebral blood 1258
flow measurement. Neuroimage 85(Pt 1), 51-63. 1259
[37] Buckley EM, Parthasarathy AB, Grant PE, Yodh AG, 1260
Franceschini MA (2014) Diffuse correlation spectroscopy 1261
for measurement of cerebral blood flow: Future prospects. 1262
Neurophotonics 1, 011009. 1263
[38] Selb J, Wu KC, Sutin J, Lin PI, Farzam P, Bechek S, Shenoy 1264
A, Patel AB, Boas DA, Franceschini MA, Rosenthal ES 1265
(2018) Prolonged monitoring of cerebral blood flow and 1266
autoregulation with diffuse correlation spectroscopy in neu- 1267
rocritical care patients. Neurophotonics 5, 045005. 1268

Uncorrected Author Proof
V. Spera et al. / Transcranial NIR on EEG 17
[39] Chai WJ, Abd Hamid AI, Abdullah JM (2018) Working
1269
memory from the psychological and neurosciences perspec-1270
tives: A review. Front Psychol 9, 401.1271
[40] Hogg RV (1987) Engineering Statistics. Macmillan Pub Co,1272
New York.1273
[41] Hollander M, Wolfe DA (1999) Nonparametric Statistical
1274
Methods, 3rd Edition. John Wiley & Sons, Inc., Hobo-1275
ken, NJ.1276
[42] Batashvili M, Staples PA, Baker I, Sheffield D (2019)1277
Exploring the relationship between gamma-band activity
1278
and maths anxiety. Cogn Emot 33, 1616-1626.1279
[43] Guragain B, Rad AB, Wang C, Verma AK, Archer L, Wil-1280
son N, Tavakolian K (2019) EEG-based classification of1281
microsleep by means of feature selection: An application
1282
in aviation. Annu Int Conf IEEE Eng Med Biol Soc 2019,1283
4060-4063.1284
[44] Keil A, Debener S, Gratton G, Junghofer M, Kappenman1285
ES, Luck SJ, Luu P, Miller GA, Yee CM (2014) Com-
1286
mittee report: Publication guidelines and recommendations1287
for studies using electroencephalography and magnetoen-
1288
cephalography. Psychophysiology 51, 1-21.1289
[45] Newson JJ, Thiagarajan TC (2018) EEG frequency bands1290
in psychiatric disorders: A review of resting state studies.1291
Front Hum Neurosci 12, 521.1292
[46] Tewarie P, Hillebrand A, van Dijk BW, Stam CJ, O’Neill
1293
GC, Van Mieghem P, Meier JM, Woolrich MW, Mor-
1294
ris PG, Brookes MJ (2016) Integrating cross-frequency
1295
and within band functional networks in resting-state
1296
MEG: A multi-layer network approach. Neuroimage 142,
1297
324-336.1298
[47] Spitzer B, Haegens S (2017) Beyond the status quo: A role1299
for beta oscillations in endogenous content (re)activation.1300
eNeuro 4, ENEURO.0170-17.2017.1301
[48] Vargas E, Barrett DW, Saucedo CL, Huang LD, Abraham
1302
JA, Tanaka H, Haley AP, Gonzalez-Lima F (2017) Bene-1303
ficial neurocognitive effects of transcranial laser in older
1304
adults. Lasers Med Sci 32, 1153-1162.
1305
[49] Wang X, Dmochowski J, Zeng L, Kallioniemi E, Husain1306
M, Gonzalez-Lima F, Liu H (2019) Transcranial photo-1307
biomodulation with infrared laser increases power of brain1308
oscillations. bioRxiv, 535757.
1309
[50] Zomorrodi R, Loheswaran G, Pushparaj A, Lim L (2019)
1310
Pulsed near infrared transcranial and intranasal photo-1311
biomodulation significantly modulates neural oscillations:
1312
A pilot exploratory study. Sci Rep 9, 6309.
1313
[51] Tallon-Baudry C, Kreiter A, Bertrand O (1999) Sustained1314
and transient oscillatory responses in the gamma and beta1315
bands in a visual short-term memory task in humans. Vis
1316
Neurosci 16, 449-459.
1317
[52] Senkowski D, Herrmann CS (2002) Effects of task difficulty
1318
on evoked gamma activity and ERPs in a visual discrimina-1319
tion task. Clin Neurophysiol 113, 1742-1753.1320
[53] Herrmann CS, Frund I, Lenz D (2010) Human gamma-band1321
activity: A reviewon cognitive and behavioral correlates and1322
network models. Neurosci Biobehav Rev 34, 981-992.1323
[54] Tallon-Baudry C, Bertrand O (1999) Oscillatory gamma
1324
activity in humans and its role in object representation.
1325
Trends Cogn Sci 3, 151-162.1326
[55] Guntekin B, Basar E (2014) A review of brain oscillations in1327
perception of faces and emotional pictures. Neuropsycholo-1328
gia 58, 33-51.1329
[56] Busch NA, Herrmann CS, Muller MM, Lenz D, Gruber
1330
T (2006) A cross-laboratory study of event-related gamma1331
activity in a standard object recognition paradigm. Neuroim-1332
age 33, 1169-1177.1333
[57] Busch NA, Debener S, Kranczioch C, Engel AK, Herrmann 1334
CS (2004) Size matters: Effects of stimulus size, duration 1335
and eccentricity on the visual gamma-band response. Clin 1336
Neurophysiol 115, 1810-1820. 1337
[58] Gruber T, Tsivilis D, Montaldi D, Muller MM (2004) 1338
Induced gamma band responses: An early marker of 1339
memory encoding and retrieval. Neuroreport 15, 1837- 1340
1841. 1341
[59] Keil A, Muller MM, Ray WJ, Gruber T, Elbert T (1999) 1342
Human gamma band activity and perception of a gestalt. J1343
Neurosci 19, 7152-7161. 1344
[60] Basar E, Demiralp T, Schurmann M, Basar-Eroglu C, 1345
Ademoglu A (1999) Oscillatory brain dynamics, wavelet 1346
analysis, and cognition. Brain Lang 66, 146-183. 1347
[61] Basar E, Basar-Eroglu C, Guntekin B, Yener GG (2013) 1348
Brain’s alpha, beta, gamma, delta, and theta oscillations in 1349
neuropsychiatric diseases: Proposal for biomarker strate- 1350
gies. Suppl Clin Neurophysiol 62, 19-54. 1351
[62] Roh D, Park S (2016) Brain multimodality monitoring: 1352
Updated perspectives. Curr Neurol Neurosci Rep 16, 56. 1353
[63] Mohammed HS (2016) Transcranial low-level infrared laser 1354
irradiation ameliorates depression induced by reserpine in 1355
rats. Lasers Med Sci 31, 1651-1656. 1356
[64] Pruitt T, Wang X, Wu A, Kallioniemi E, Husain MM, 1357
Liu H (2020) Transcranial photobiomodulation (tPBM) 1358
with 1,064-nm laser to improve cerebral metabolism of the 1359
human brain in vivo.Lasers Surg Med 52, 807-813. 1360
[65] Chao LL (2019) Effects of home photobiomodulation 1361
treatments on cognitive and behavioral function, cere- 1362
bral perfusion, and resting-state functional connectivity in 1363
patients with dementia: A pilot trial. Photobiomodul Pho- 1364
tomed Laser Surg 37, 133-141. 1365
[66] Holmes E, Barrett DW, Saucedo CL, O’Connor P, Liu 1366
H, Gonzalez-Lima F (2019) Cognitive enhancement by 1367
transcranial photobiomodulation is associated with cere- 1368
brovascular oxygenation of the prefrontal cortex. Front 1369
Neurosci 13, 1129. 1370
[67] Caldieraro MA, Cassano P (2019) Transcranial and sys- 1371
temic photobiomodulation for major depressive disorder: 1372
A systematic review of efficacy, tolerability and biological 1373
mechanisms. J Affect Disord 243, 262-273. 1374
[68] Song X, Hu W, Yu H, Wang H, Zhao Y, Korngold R, Zhao 1375
Y (2020) Existence of circulating mitochondria in human 1376
and animal peripheral blood. Int J Mol Sci 21, 2122. 1377
[69] Del Felice A, Castiglia L, Formaggio E, Cattelan M, Scarpa 1378
B, Manganotti P, Tenconi E, Masiero S (2019) Personal- 1379
ized transcranial alternating current stimulation (tACS) and 1380
physical therapy to treat motor and cognitive symptoms in 1381
Parkinson’s disease: A randomized cross-over trial. Neu- 1382
roimage Clin 22, 101768. 1383
[70] Kim J, Jang KI, Roh D, Kim H, Kim DH (2020) A direct 1384
comparison of the electrophysiological effects of transcra- 1385
nial direct and alternating current stimulation in healthy 1386
subjects. Brain Res 1747, 147065. 1387
[71] Guleyupoglu B, Schestatsky P, Edwards D, FregniF, Bikson 1388
M (2013) Classification of methods in transcranial electri- 1389
cal stimulation (tES) and evolving strategy from historical 1390
approaches to contemporary innovations. J Neurosci Meth- 1391
ods 219, 297-311. 1392
[72] Sharma SK, Kharkwal GB, Sajo M, Huang YY, De Taboada 1393
L, McCarthy T, Hamblin MR (2011) Dose response effects 1394
of 810nm laser light on mouse primary cortical neurons. 1395
Lasers Surg Med 43, 851-859. 1396
[73] Cassano P, Tran AP, Katnani H, Bleier BS, Hamblin MR, 1397
Yuan Y, Fang Q (2019) Selective photobiomodulation for 1398

Uncorrected Author Proof
18 V. Spera et al. / Transcranial NIR on EEG
emotion regulation: Model-based dosimetry study. Neu-
1399
rophotonics 6, 015004.1400
[74] Yuan Y, Cassano P, Pias M, Fang Q (2020) Transcranialpho-1401
tobiomodulation with near-infrared light from childhood1402
to elderliness: Simulation of dosimetry. Neurophotonics 7,1403
015009.
1404
[75] Tedford CE, DeLapp S, Jacques S, Anders J (2015) Quan-1405
titative analysis of transcranial and intraparenchymal light1406
penetration in human cadaver brain tissue. Lasers Surg Med1407
47, 312-322.
1408
[76] Lychagov VV, Tuchin VV, Vilensky MA, Reznik BN, Ichim 1409
T, De Taboada L (2006) Experimental study of NIR trans- 1410
mittance of the human skull. Proceedings Volume 6085, 1411
Complex Dynamics and Fluctuations in Biomedical Photon- 1412
ics III; 60850T (2006) https://doi.org/10.1117/12.650116. 1413