Developmental changes in neonatal hemodynamics during tactile stimulation using whole-head functional near-infrared spectroscopy

Neural-activity-associated hemodynamic changes have been used to noninvasively measure brain function in the early developmental stages. However, the temporal changes in their hemodynamics are not always consistent with adults. Studies have not evaluated developmental changes for a long period using the same stimuli; therefore, this study examined the normalized relative changes in oxygenated hemoglobin ( Δ [oxy-Hb]) in full-term infants and compared them with neonates up to 10 months of age during the administration of tactile vibration stimuli to their limbs using whole-head functional near-infrared spectroscopy. The time to peak of normalized Δ [oxy-Hb] was not affected by age. The amplitude of normalized Δ [oxy-Hb] showed an effect of age in broader areas, including sensorimotor-related but excluding supplementary motor area; the amplitude of normalized Δ [oxy-Hb] decreased the most in the 1 – 2-month-old group and later increased with development. We hypothesized that these results may reflect developmental changes in neural activity, vasculature, and blood oxygenation.


Introduction
Cerebral hemodynamic responses are interpreted as indirect indicators of underlying neural activity and widely applied in noninvasive brain function measurement techniques.Neurovascular signals stimulate blood delivery higher than required by neural tissues in the adult brain, causing a localized increase in oxygenated hemoglobin and a decrease in deoxygenated hemoglobin.
Typical hemodynamic changes like in adults, i.e., an increase in the relative changes in concentration of oxygenated hemoglobin (Δ[oxy-Hb]) and a decrease in the relative changes in concentration of deoxygenated hemoglobin (Δ[deoxy-Hb]), and positive BOLD signals have been detected from fetal to neonatal stages.Conversely, several studies on neonates and preterm infants have reported changes in the hemoglobin pattern different from adults (Anderson et al., 2001;Kusaka et al., 2004;Sakatani et al., 1999).Anderson et al. evaluated the bilateral superior temporal gyri during auditory stimulation in preterm and full-term infants and observed brain activity in 14/20 infants; of 14 infants, 9 had a negative BOLD signal (Anderson et al., 2001).Sakatani et al. reported that 60.7 % of neonates under study showed an increase in Δ[oxy-Hb], Δ[deoxy-Hb], and Δ[total-Hb] (Δ[oxy-Hb] + Δ[deoxy-Hb]), whereas 32.1 % exhibited an increase in Δ[oxy-Hb] and Δ[total-Hb] and a decrease in Δ[deoxy-Hb] in the bilateral frontal regions during auditory stimulation (Sakatani et al., 1999).Moreover, Kusaka et al. reported that preterm infants showed reductions and elevations in Δ[oxy-Hb] and Δ [deoxy-Hb] in the visual cortex during a stroboscopic white flashing light projection onto the eyelids at 8 Hz frequency (Kusaka et al., 2004).
Studies on neural-activity hemodynamics have reported that development affects changes in the amount and latency of hemoglobin (Arichi et al., 2012;Colonnese et al., 2008;Kozberg et al., 2013;Yamada et al., 1997Yamada et al., , 2000)).Rats aged 12-13 days showed a decrease in Δ[oxy-Hb] and an increase in Δ[deoxy-Hb]; however, with growth, Δ[oxy-Hb] increased and Δ[deoxy-Hb] decreased, similar to adults (Kozberg et al., 2013).A functional MRI study assessing BOLD signal changes throughout human development found differences in the signal between preterm infants, infants with a term-equivalent age, and adults (Arichi et al., 2012).Reportedly, the time to peak decreases with maturation between approximately 32 and 46 weeks of postmenstrual age; the positive peak amplitude increases with maturation.A positive BOLD signal was reported from early development to 7 weeks of age, whereas a negative BOLD signal was detected in a calcarine fissure after 8 weeks of age during visual stimulation (Yamada et al., 1997(Yamada et al., , 2000)).Colonnese et al. reported that rats aged 10-12 days did not show significant BOLD signal changes, whereas rats aged ≥13 days showed a significant BOLD signal in the somatosensory cortex (Colonnese et al., 2008).They found that with advancing age, the proportion of rats showing a positive BOLD response increased, accompanied by an increasing amplitude and decreasing time to peak in the averaged BOLD time curves.
In this study, we focused on the developmental changes in Δ[oxy-Hb] because its rate of change is larger than Δ[deoxy-Hb] and shows various temporal changes.Functional NIRS studies have shown that Δ[deoxy-Hb] has smaller changes than Δ[oxy-Hb], particularly in the early developmental period when hemoglobin changes are smaller than adults, and Δ[deoxy-Hb] is even smaller.Although increasing the number of trials can enhance the detection of minute signals, a study design demanding extended measurement durations is unfeasible for infants in early development.Therefore, Δ[deoxy-Hb] in the early stages of development is an unsuitable indicator of brain activity because the signal changes are so small that detecting signal changes at individual and group levels may not be possible.The second reason is that studies have reported more diversity in peak times and subsequent changes in Δ[oxy-Hb] and Δ[total-Hb] than Δ[deoxy-Hb] (Chen et al., 2014;Kozberg et al., 2013).In our study, we assessed Δ[oxy-Hb] waveforms because Δ[total-Hb] and Δ[oxy-Hb] waveforms show almost the same temporal changes.To achieve this, we normalized Δ[oxy-Hb] and focused on developmental changes in time to peak and variability of each time window.Most studies have evaluated developmental changes up to 2 months of age, and no studies with an adequate number of infants have evaluated developmental changes in the hemodynamic temporal changes over a longer period using the same stimuli.Herein, we examined Δ[oxy-Hb] in 60 full-term infants, from birth to approximately 10 months of chronological age (CA), during the delivery of tactile vibratory stimuli using a functional NIRS system with channels.

Participants
In this study, 62 full-term infants (24 females and 38 males) were included.Two infants were excluded during data processing because of low data quality, based on the analysis criteria; therefore, 60 full-term infants (23 females and 37 males) with a mean gestational age of 39.3 weeks (38-41.3weeks) were evaluated.Infants displayed no anatomical abnormalities and were divided into five groups according to their CA (Table 1): neonates, 1-2, 2-4, 4-6, and ≥6 months.

Ethics
The Ethics Committees of Tokyo Metropolitan University (H3-38) and Kagawa University Hospital, Kagawa, Japan (H27-130), approved this study.Caregivers of infants provided informed consent.They were informed that they had the option to withdraw their children from the study at any point.The caregivers received no payment for their participation in the study.

Data acquisition
A functional NIRS system with 91 measurement channels (ETG-7000; Hitachi, Ltd., Tokyo, Japan) was used for data collection.All signals were simultaneously measured and stored within a 100-ms sampling interval.Two near-infrared laser diode light sources were used, with 785 and 830 nm wavelengths.The signals at the 91 positions were measured using the following configuration: 30 irradiation and detection positions (Figure S1).Avalanche photodiodes positioned at approximately 21-30 mm were used to detect the reflected light from the emitters.
The probes of functional NIRS system were arranged at uniform intervals and placed on the entire scalp (Figure S1).The center of the lowest measurement position line in the frontal region was adjusted to coincide with Fpz, according to the international 10-10 system for electroencephalography.The center of the lowest measurement position line in the temporal region was adjusted to coincide with T7 and T8 scalp sites, and the center of probe holder in the occipital region was adjusted to coincide with Oz.The remaining detectors and emitters were arranged at uniform intervals.
To estimate that the measurement channels corresponded to the approximate anatomical brain regions of infants, coordinates of probes and a four-dimensional optical head model for full-term infants (Brigadoi et al., 2014) were imported into AtlasViewer, an open-source software package based on MATLAB (MathWorks, Inc., Natick, MA, USA) (Table S1).
Each infant was encouraged to sleep by feeding milk 30 min before the measurement and held by a pediatrician for measurement during natural sleep.In the tactile task, four tiny vibration motors (VM-412; STL Co., Ltd., Tokorozawa, Japan), covered with an insulation cap (TIC-14; Nichifu Co., Ltd., Osaka, Japan) and fixed with surgical tape (SGM 25; Nichiban Co., Ltd., Osaka, Japan) to the infant's arms and legs, were used to present vibration stimuli.Each motor was switched on for 10 s, followed by a 20-25-s rest period.The infant's palms and soles were stimulated six times in a pseudo-random and independent manner.During the 10-s stimulation period, continuous and intermittent vibrations were alternatively presented.Functional recording of NIRS was performed in a quiet room in a maternity unit.
The infant's behavior and facial movements were recorded on a check sheet.Two digital video cameras (GoPro HERO3+; GoPro, Inc., CA, USA) were used to record the movement of each infant during measurements to remove motion artifacts during post-signal preprocessing.

Signal preprocessing
Signal preprocessing analysis, focusing on Δ[oxy-Hb], was performed using modified Beer-Lambert law (Maki et al., 1995).Δ[oxy-Hb] data were analyzed using MATLAB-based Platform for Optical Topography Analysis Tools.
After band-pass filtering of Δ[oxy-Hb] at 0.01-0.8Hz, we flagged Δ[oxy-Hb] as unstable signal artifacts, defined as changes of >0.1 mM⋅mm between two consecutive samples.This band-pass filter was applied only for artifact detection and had no impact on the subsequent analysis.To remove task-design-induced components and measurementsystem-derived variations, Δ[oxy-Hb] for individual channels were digitally high-pass-filtered at 0.01 Hz.The data were smoothed using a 3 s moving average to remove the effects of heart rate.
The fitting used mean of the 5-s period from − 5 to 0 s before the task period onset.Any blocks, including the flags defined as artifacts, in the first step were removed.Any channels containing artifacts in >8/24 blocks were excluded from the single-participant level.Infants that contained artifacts in >8/24 blocks in all channels were excluded from the analysis.Two infants were excluded based on this criterion.Table 1 shows the data obtained after exclusion.
To evaluate the temporal change in cerebral hemodynamics, Δ[oxy-Hb] at each measurement position (channel) were normalized in each infant: where Δ[Hb](u) represents Δ[oxy-Hb] at time u, and T represents the length of the block.We plotted the average waveforms for each age group by assessing normalized Δ[oxy-Hb] in response to all tactile vibration stimuli.
Supplemental Results show the one-sample t-test results of the mean values obtained 5-15 s after the onset of stimuli, shown as t-maps (p < 0.05, uncorrected), when each sensory stimulus was presented to each age group (Figure S2).Subsequently, we evaluated the time to peak from stimulation onset and the mean amplitude of normalized Δ[oxy-Hb] based on the time window (0-5, 5-10, 10-15, and 15-25 s).Additionally, a 10-s moving average was applied to exclude small fluctuations in seven brain regions of interest (ROIs) in each infant, namely, middle frontal gyrus (Fron-tal_Mid), paracentral lobule (Paracentral_Lobule), postcentral gyrus (Postcentral), precentral gyrus (Precentral), superior parietal gyrus (Parietal_Sup), superior frontal gyrus (Frontal_Sup), and supplementary motor area (Supp_Motor_Area).The measurement positions included in the seven ROIs were selected for both hemispheres (Table S1).

Statistical analysis
Statistical analyses were performed using Statistical Package for the Social Sciences (IBM Corp., Armonk, NY, USA).One-way analysis of variance (ANOVA) was used to determine whether statistically significant differences existed between the peak time and age groups in each region.
Moreover, two-way mixed ANOVA was used to investigate the effects of time window as within-subject factor, age groups as between-subject factor, and interaction between time window and age group on mean normalized Δ[oxy-Hb] amplitude for each region.To test the sphericity hypothesis, Mauchly's sphericity test was used.In case the hypothesis of sphericity (p < 0.05) is rejected, the analyses were based on the results of the multivariate Greenhouse-Geisser test.Bonferroni test was used to correct for rejection in the post hoc test, with 0.05 threshold.

Data and code availability
Due to the absence of ethical committee approval for data sharing agreement, we cannot share any individual-level behavioral and imaging data used in this study.MATLAB scripts for stimulus presentation and data processing are available from Y.F upon request.

Temporal changes in cerebral hemodynamics
Temporal changes in normalized Δ[oxy-Hb] in the postcentral gyrus tended to peak 6-9 s after the onset of tactile stimulation in each age group (Fig. 1).
In the neonate group, normalized Δ[oxy-Hb] began to increase 2-3 s after stimulation onset and peaked within approximately 6.6 s.Moreover, normalized Δ[oxy-Hb] did not persist at maximum; rather, they quickly decreased to a lower level than recorded at baseline, reaching a minimum value approximately 18.4 s later, and then increasing and returning to the baseline level.
In the 1-2-month-old group, normalized Δ[oxy-Hb] decreased temporarily after stimulation onset and reached a minimum value approximately 2.7 s later, after which normalized Δ[oxy-Hb] level increased and peaked approximately 6.0 s after stimulation onset.normalized Δ[oxy-Hb] decreased to a level below that recorded at baseline, reached a minimum value approximately 18.8 s later, and then increased and returned to the baseline level.
In the 2-4-month-old group, normalized Δ[oxy-Hb] began to increase 2-3 s after stimulation onset and peaked within approximately 6.0 s.Subsequently, they slowly decreased to the baseline level, with a small second peak recorded at 17.0 s after stimulation onset.
In the 4-6-month-old group, normalized Δ[oxy-Hb] began to increase 2-3 s after stimulation onset and reached a relatively modest peak within approximately 8.2 s.Then, they formed a second peak 15.4 s after stimulation onset and then returned to the baseline level.
In the ≥6-month-old group, normalized Δ[oxy-Hb] began to increase 2-3 s after stimulation onset and peaked within approximately 6.4 s.Subsequently, normalized Δ[oxy-Hb] reached a second peak after approximately 16.3 s and then returned to the baseline level.normalized Δ[oxy-Hb] recorded in each region are shown in the Supporting Information (Figure S3).

Peak time of normalized Δ[oxy-Hb]
To statistically evaluate the peak time of normalized Δ[oxy-Hb], one-way ANOVA was performed in each brain region.No significant differences in any region were observed among the age groups (Table 2, Figure S4).

Mean normalized Δ[oxy-Hb] amplitude in each time window
Table 3 presents the results of the two-way mixed ANOVA performed to investigate the effects of the time window as within-subject factor, age groups as between-subject factor, and interaction between time window and age group on the mean normalized Δ[oxy-Hb] amplitude in each brain region.No interaction was observed between the time window and age group in any region examined in this study (Table 3).
The main effect of the time window was statistically significant in all regions (p < 0.05) (Table 3).The mean values for each time window exhibited a common trend in each region, although at different statistical significance levels, in the postcentral gyrus, significant differences were observed between 0-5 and 5-10 s (p < 0.0005), 5-10 and 15-25 s (p < 0.005), and 10-15 and 15-25 s time windows (p < 0.05) (Fig. 2).
The main effect of age groups was significant in sensorimotor areas other than the Supp_Motor_Area (Table 3).This statistic was highest in the postcentral gyrus ([F(4, 55) = 5.06, p < 0.005, η 2 p = 0.27]).The mean values for each age group exhibited a common trend in each region; although at different statistical significance levels (Fig. 3), the mean normalized Δ[oxy-Hb] values for the neonate group were positive, whereas those for the 1-2-month-old group were the lowest and negative, increasing with age.In the postcentral gyrus, significant differences were observed between neonate and 1-2-month-old groups (p < 0.05), 1-2-and 2-4-month-old (p < 0.05), 1-2-and 4-6-month-old (p < 0.005), and 1-2-and ≥6-month-old groups (p < 0.005).Although the statistical significance of each region varied, the same trend was observed in almost all regions.

Discussion
Temporal changes in cerebral hemodynamics related to stimulation during early development do not necessarily correspond to those observed in adults but are diverse.To examine sustained developmental alterations using identical tactile stimuli, we recorded hemodynamic variations in full-term infants, spanning from neonates to those aged 10 months, during the administration of tactile vibration stimuli to their limbs.Our results showed that the peak time from stimulus onset was not affected by age in any brain region.While evaluating the mean amplitude of normalized Δ[oxy-Hb], the main effect of age was significant in broader areas, including sensorimotor-related but excluding    [oxy-Hb] was the smallest and negative in the 1-2-month-old group and later increased with development.We speculated that these results reflect developmental changes in neural activity, vasculature, and blood oxygenation.

Developmental effect on peak time
The common feature of hemodynamic responses to tactile vibration stimulation was that the peak occurred 6-10 s after stimulation onset and then returned to the baseline level; when the peak time was compared among age groups, no statistically significant differences were observed in any region (Table 2, Figure S3).Our results differ from previous studies (Arichi et al., 2012;Colonnese et al., 2008); the time to peak did not decrease with development.We believe that this was possibly because of the difference between the development period covered in our and previous studies.and infants with an average gestational age of 41 weeks and 1 day in humans.Colonnese et al. evaluated developmental changes in rats, focusing on postnatal day (P) 10 up to adulthood.They estimated that this period in rats corresponds to 28 weeks of gestational age to 7-8 years in humans, according to a study comparing developmental timing across species in mammals (Clancy et al., 2001).Based on the literature and our study, the peak time may decrease in periods corresponding to the human fetal period but may not change after birth.
Two factors affecting time to peak were considered: capillaries/ penetrating arterioles and the "fast" mechanism of retrograde propagation through the endothelium of pial arterioles.Kozberg et al. (2013) reported that neurovascular coupling is localized in capillaries/penetrating arterioles in younger rats whereas hyperemia in capillaries/penetrating arterioles and pial arterioles in adult rats.In rodent studies, the capillary system is reconstituted after birth, and the density of the capillaries located in the gray matter of P20 animals is twice than recorded in the immediate postnatal period because of continuous angiogenesis and degeneration (Coelho-Santos and Shih, 2020).Various species may exhibit differences in developmental timing; however, developmental changes in capillaries/penetrating arterioles may affect the peak time.
The other factor affecting the peak time may be the developmental changes in vasodilation on the "fast" mechanism of retrograde propagation through the endothelium of pial arterioles.In adult rats, exposure to tactile stimuli induced hyperemia in capillaries/penetrating arterioles and pial arterioles (Kozberg et al., 2013).Chen et al. (2014) reported that dilation of pial arterioles is caused by retrograde propagation by endothelium in pial arterioles during somatosensory activation.They found that when adult rats were exposed to 12 s of somatosensory stimulation, hemoglobin responses in ROI formed a 4-6-s peak and 10-12-s plateau, with only a 4-6-s peak observed in the distal pial arterioles.Furthermore, wide-field endothelial disruption in the pial arterioles attenuated the hemoglobin response; it was attenuated at peak and plateau, with peak attenuation being more pronounced.The mechanism of endothelium-dependent vasodilation of pial arterioles in the brain is not well understood; however, the underlying mechanism in the peripheral vasculature has been investigated.Tallini et al. reported that endothelium dilates via an electrical retrograde "transient fast distal process" and a bidirectional "sustained slow proximal process" caused by calcium waves in arterioles of the cremaster muscle in mice (Tallini et al., 2007).The two phases of endothelial vasodilation regulation in mouse cremaster muscle arterioles might apply to pial arteriolar dilation in the brain (Chen et al., 2014).If a similar mechanism is at work in the human brain, the developmental changes in the "fast" phase of vasodilation control in endothelium can affect the peak time.

Developmental effect on amplitude of normalized Δ[oxy-Hb]
We detected a main effect of the time window in all examined regions, and the post hoc test showed significant differences between common time windows, even though the significance of the differences varied region to region (Table 3, Fig. 3).Although the occurrence of the main effect of time window was likely in any region because it is normalized for each channel, the hierarchical processing of somatosensory perception from lower to higher levels is dynamically developed in the perinatal period (Whitehead et al., 2019), which is consistent with our observation of significant hemodynamic changes in response to tactile stimuli over a broad region (Figure S2).
Furthermore, the main effect of age was significant in broad areas (p < 0.05), including sensorimotor-related but excluding supplementary motor area (Table 3).The main effect of age was highest in the postcentral gyrus ([F(4, 55) = 5.06, p < 0.005, η 2 p = 0.27]).Post hoc tests showed that mean normalized Δ[oxy-Hb] was positive in the neonatal group but was lowest and negative in the 1-2-month-old group; mean normalized Δ[oxy-Hb] in the postcentral gyrus increased with age (Fig. 3).These trends were similar in most regions.Studies have reported that the positive BOLD signal associated with visual cortex stimulation becomes negative in infants after 8 weeks (Yamada et al., 1997(Yamada et al., , 2000)).These results are similar to the transient decrease in Δ[oxy-Hb] at 1-2 months in our study, although ROIs differed and the timing was different.
These results could be influenced by developmental changes in two factors: physiological anemia (hemoglobin concentration) and cerebral blood flow (CBF).In physiological anemia, hemoglobin concentrations decrease temporally during the first 2-3 months of life (Dallman et al., 1980;Lundstrom et al., 1977;Saarinen and Siimes, 1978).In full-term infants, the rise in oxygen levels due to regular breathing, following birth leads to a swift surge in tissue oxygenation, leads to the inhibition of erythropoietin production and erythropoiesis.This decrease in erythropoiesis and the short life span of neonatal erythrocytes decrease the concentration of hemoglobin during the first 2-3 months of life.Then, hemoglobin concentration stays stable over several weeks and gradually increases with the stimulation of erythropoietin at 4-6 months.In adult studies, the absolute hemoglobin concentration and hematocrit level affect BOLD signal, as previously assessed using a task-based paradigm (Gustard et al., 2003;Levin et al., 2001) and intrinsic functional connectivity (Ward et al., 2020;Yang et al., 2014).
The second factor may be the developmental changes in CBF.Wong et al. reported that CBF increased with development until 12.5 months of CA, plateaued between 12.5-30 months, and increased as growth progressed thereafter (Wong et al., 2019).Furthermore, in a study by Hirata et al. using single-photon emission computed tomography, CBF increased with growth up to 12 months after birth (Hirata et al., 2018).The lowest amplitude and negative values in the 1-2-month-old group may be related to the period when oxygen supply is reduced due to low CBF and the lowest hemoglobin concentration due to physiologic anemia.
We observed various temporal changes at all ages on an individual level.Although typical increases in Δ[oxy-Hb] were observed in adults, we noted decreases and no apparent changes with tactile stimulation (Figure S5).Moreover, we did not always observe common characteristics of temporal changes in individuals among infants who were longitudinally measured (Figure S6).Instead, we observed a range of temporal changes among individuals of the same age.These variations cannot be fully explained by developmental changes in the structure and function of brain networks within and between individuals.Rather, they may be attributed to individual developmental differences in at least three factors related to neurovascular coupling: hemoglobin concentration, CBF, and vasodilation.Hemoglobin concentrations are reportedly affected by internal and external environmental conditions of an individual, such as preterm birth (Dallman et al., 1980;le Cessie et al., 2002;Lundstrom et al., 1977), nutritional intake (e.g., iron, folate, and vitamin B12 supplementation) (Lundstrom et al., 1977;Stevens et al., 2013), infections (e.g., malaria and human immunodeficiency virus) (le Cessie et al., 2002;Stevens et al., 2013), and national economic situation (Stevens et al., 2013).Although cerebral blood vessels are morphologically homogeneous, single-cell RNA sequencing technology has revealed that gene expression differs among regions, and gradations along the arteriovenous axis (zonation) exist (Saunders et al., 2018;Schaeffer and Iadecola, 2021;Vanlandewijck et al., 2018).Individual differences in the region and timing of gene expression may affect vasodilation.
The main effect of age on the supplementary motor area was insignificant ([F(4, 55) = 1.77, p = 0.15, η 2 p = 0.11]), probably because the stimuli used in this study were only passive tactile vibration stimuli.Studies have reported that the supplementary motor area is critical to voluntary motor initiation (Mushiake et al., 1990(Mushiake et al., , 1991) ) and control (Roland et al., 1980;Sadato et al., 1997).In our study, some infants were temporarily induced to move by tactile vibration stimuli; however, the stimuli used were not tasks that required voluntary movement.Therefore, we speculate that age has no major effect on the supplementary motor area.

Limitations of the study
Despite the presentation of stimuli in functional neuroimaging studies, brain hemodynamic changes may not be observed in the postnatal period (Colonnese et al., 2008).When this phenomenon occurs, it is impossible to determine whether there is no neural activity in response to a specific stimulus in ROI, or there is neural activity but no statistically significant hemodynamic change.When tactile vibration stimuli were applied to each limb in our study, typical increases in Δ[oxy-Hb] were not always observed in the sensorimotor area (precentral and postcentral gyrus).T-maps (p < 0.05, uncorrected) of oxygenated hemoglobin changes during tactile vibration stimulation of each limb (Figure S2) showed that no significant increase or decrease in Δ[oxy-Hb] was always observed in younger infants.However, even in such cases, for example, when tactile vibration stimulation was administered to the right foot in the 2-4-month-old group, decreases in Δ[oxy-Hb] in the prefrontal region were observed.Unlikely, hemodynamic changes occur only in the prefrontal region without primary sensorimotor information processing because the hierarchical processing of somatosensory perception in the perinatal period is dynamically developing (Whitehead et al., 2019).The interpretation of this phenomenon might be the effect of reduced oxygen supply in all regions of the brain because of reduced hemoglobin concentration, CBF, and differences in oxygen consumption per brain region.To address this issue, simultaneously evaluating oxygen supply/demand, electrophysiological and hemodynamic brain characteristics, developmental changes in the hemodynamic time course, and differences between regions is necessary.

Future prospects
Our results revealed that temporal changes in Δ[oxy-Hb] related to tactile stimulation change with development, and we speculate that these changes are influenced by developmental changes in three factors-pial arterioles, hemoglobin concentration, and CBF-rather than neural activity.Clarifying the relationship between the three factors and hemodynamics and constructing a model of hemodynamic temporal changes using these three factors in future studies is necessary.

Conclusion
We evaluated temporal changes in cerebral hemodynamics during tactile stimulation in full-term infants, from neonates to 10 months of CA, and found that these changes were influenced by development in broader areas, including sensorimotor-related but excluding supplementary motor area.This developmental change was not a linear increase but was lowest at 1-2 months, probably due to the reduced oxygen supply as a result of physiological anemia and lower CBF.Future studies should simultaneously evaluate the relationship between hemodynamic temporal changes and the structure and function of neurovascular units and physiological factors at the individual level.

Fig. 1 .
Fig. 1.Temporal changes in normalized Δ[oxy-Hb] during sensory stimulation of postcentral gyrus The solid lines represent the temporal changes in normalized Δ[oxy-Hb] during tactile stimulation in different groups; orange line with circle markers represent the neonate group; yellow line with triangle markers represents the 1-2-monthold group; green line with square markers represents the 2-4-month-old group; light blue line with diamond markers represents the 4-6-month-old group; purple line with star markers represents the ≥6-month-old group.Tactile vibration stimuli were presented for 0-10 s, (in gray).Error bars represent standard error.

Table 1
Demographic information.

Table 2
Results of one-way ANOVA performed to analyze the effect of age on the time to peak in each brain region of interest.

Table 3
Results of two-way mixed ANOVA performed to analyze the effect of amplitude of the time window, age groups, and interaction between time window and age group in each region.
Results of the two-way mixed ANOVA with the time window (0-5, 5-10, 10-15, and 15-25 s) as a within-subject factor and age group (neonates and 1-2, 2-4, 4-6, and ≥6 months of age) as a between-subject factor in each region of interest.*p< 0.05 ** p < 0.001 *** p < 0.0005 n.s.not significant.supplementarymotor area.The mean values for each age group exhibited a common trend in each region: amplitude of normalized Δ