2.1.

Subjects

The study was carried out with 134 healthy subjects (85 female, 49 male, age $24.7\pm 3.4$ years, and range 20 to 46 years). The subjects were all right-handed, according to the Edinburgh Handedness Inventory.²³ Subjects were nonsmoking and indicated neither current nor previous history of neurological and psychiatric disorders or alcohol and drug abuse. Subjects were asked to refrain from consuming caffeine and eating 2 h prior to the experiment. The study protocol was approved by the Ethics Committee of the Canton of Bern. Informed consent was obtained from all subjects before the measurements. Subjects were also informed of their right to discontinue participation at any time.

2.2.

Experimental Protocol

The resting state data were taken from a set of studies with different stimuli. Each measurement began with a baseline phase lasting 8 min, during which the subjects sat upright in a comfortable chair in a dark room. For this study, we examined only the last 5 min of this baseline period. Each subject was measured on four different days but at the same time of day to prevent chronobiological artifacts. They were asked to keep their eyes open throughout the entire measurement and to move their head or body as little as possible during the measurement to avoid movement artifacts. Additionally, the subjects were asked to fill out two questionnaires before and after each measurement in order to assess their mood: the positive affect negative affect schedule (PANAS)²⁴ and the self-assessment manikin test (SAM; five points scale).²⁵ The PANAS and SAM questionnaires are used as tools to measure state influence. Trait influence evaluation was not performed in our study. Additionally, we determined the chronotype by the Horne and Östberg morningness-eveningness questionnaire.²⁶ Measurements were performed between 7:00 am and 9:00 pm. The mean room temperature was $22.8°\mathrm{C}\pm 0.6°\mathrm{C}$.

2.3.

Measurement Setup

The Imagent (ISS Inc., Champaign, Illinois, USA), a multichannel FD-NIRS system, which employs a multidistance approach, was used to determine absolute values of the $[{\mathrm{O}}_2\mathrm{Hb}]$, [HHb], [tHb], and ${\mathrm{StO}}_2$ at a sampling rate of 2.5 Hz on the PFC. The Imagent’s light source consists of 16 laser diodes at 760 nm and 16 laser diodes at 830 nm. Four highly sensitive photomultiplier tubes serve as detectors. The sensors were placed bilaterally on the left and right prefrontal cortex (L-PFC and R-PFC) of subjects at position Fp1 and Fp2, according to the international 10 to 20 system.²⁷ Each of the two ISS sensors had four light emitters and one light detector connected to an optical fiber delivering the light to the photomultiplier tube. The source–detector separations ($d$) were $\sim 2.0$, 2.5, 3.5, and 4.0 cm with the sources and detectors arranged collinearly.

HR was measured by SOMNOtouch™ NIBP (SOMNOmedics GmbH, Randersacker, Germany) with a sampling rate of 4 Hz. This device calculated the HR from the ECG data by calculating the $R-R$ intervals. RR and end-tidal carbon dioxide (${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$) were measured noninvasively by a NONIN LifeSense (NONIN Medical, Plymouth, Minnesota, USA). Data were recorded at a sampling rate of 1 Hz. All data were recorded simultaneously.

2.4.

Signal Processing and Statistical Analysis

One subject was excluded from data analysis due to the perceived discomfort of the fNIRS sensors. 126 subjects completed all four measurements; only for seven subjects, the number of experimental sessions was lower. Therefore, the entire data for the current analysis comprised 518 single measurements. All signal processing was performed in MATLAB (R2017a, MathWorks, Inc., Massachusetts, USA).

3.1.

Cerebral Oxygenation and Perfusion: Higher StO₂ of Right PFC

For ${\mathrm{StO}}_2$, we found right-dominant activity ($\mathrm{\Delta }{\mathrm{StO}}_2>0$), i.e., a highly significant ($p<0.0001$) FCOA was detected in the resting state [Fig. 1(a)]. No significant ($p=0.324$) asymmetry was found for [tHb] [Fig. 1(b)]. The intersubject mean value representing the normal value of ${\mathrm{StO}}_2$ and [tHb] was ($\text{mean}\pm \mathrm{SD}$) $73.0\%\pm 5.9\%$ and $41.4\pm 9.3\mu \mathrm{M}$, respectively [Figs. 1(c) and 1(d)]. For the right and left PFC the normal values of ${\mathrm{StO}}_2$ were $73.7\%\pm 6.9\%$ (right) and $72.3\%\pm 6.1\%$ (left), and of [tHb] $41.4\pm 10.8$ (right) and $41.4\pm 10.1$ (left) [Figs. 2(a) and 2(b)]. The absolute values of ${\mathrm{StO}}_2$ and [tHb] during the resting state are shown in Figs. 2(c) and 2(d). All data were normally distributed. From this point on, the outliers were removed and not considered for subsequent evaluations. After removal of the outliers, FCOA remained highly significant ($p<0.0001$) [nonsignificant for [tHb] ($p=0.218$)].

Fig. 1

(a) FCOA of ${\mathrm{StO}}_2$ at the PFC sorted in descending order on the individual subjects. (b) Asymmetry of [tHb] and absolute values of (c) ${\mathrm{StO}}_2$, (d) [tHb], (e) ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$, (f) HR, (g) RR, and (h) PRQ displayed according to the $\mathrm{\Delta }{\mathrm{StO}}_2$ sorting, at the PFC for individual subjects during resting state. The median and the IQR are shown for each subject.

Fig. 2

Diamond box plots showing distributions of absolute (a) ${\mathrm{StO}}_2$ and (b) [tHb] values at the R-PFC and L-PFC, (c) ${\mathrm{StO}}_2$, (d) [tHb], (e) ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$, (f) $\mathrm{\Delta }{\mathrm{StO}}_2$, (g) $\mathrm{\Delta }\mathrm{tHb}$, (h) HR, (i) RR, and (j) PRQ values in resting state. The diamond spans the first quartile to the third quartile (IQR). A segment inside the diamond shows the median and whiskers above and below the box plots represent the 95% prediction interval. The asterisks indicate the level of high significance between absolute ${\mathrm{StO}}_2$ values of the R-PFC and L-PFC (*$p<0.001$, Wilcoxon signed-rank test). Outliers are not displayed.

3.2.

Cardiorespiratory Activity

Figures 1(e)–1(h) and 2(e), 2(h)–2(j) show the absolute values of ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$, HR, RR, and PRQ for the individual subjects during resting state. On the group level, mean absolute and SD values of cardiorespiratory parameters were as follows (data were normally distributed): ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$: $39.2\pm 4.4\mathrm{mmHg}$, HR: $68\pm 11$ beats/min (BPM), RR: $16.5\pm 2.9$ breaths/min (BrPM), and PRQ: $4.2\pm 0.9$. These values were all in the normal range for healthy adults at rest.

3.3.

Relationships with Systemic Physiology

A correlation matrix of $\mathrm{\Delta }{\mathrm{StO}}_2$, $\mathrm{\Delta }\mathrm{tHb}$, ${\mathrm{StO}}_2$, [tHb], ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$, HR, RR, and log(PRQ) variables during the resting state is depicted in Fig. 3(a). In detail, Table S1 in the Supplementary Material shows curve fitted models, goodness-of-fit results, BSE parameter, and a significance level for each pair of variables. Statistically significant correlations were found between cerebral and cardiorespiratory parameters for six pairs of variables: $\mathrm{\Delta }{\mathrm{StO}}_2$ versus RR ($p_{\mathrm{FDR}}=0.022$, $p=0.010$), $\mathrm{\Delta }{\mathrm{StO}}_2$ versus log(PRQ) ($p_{\mathrm{FDR}}=0.024$, $p=0.012$), ${\mathrm{StO}}_2$ versus ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$ ($p_{\mathrm{FDR}}<0.0001$, $p<0.0001$), [tHb] versus HR ($p_{\mathrm{FDR}}=0.001$, $p<0.0004$), [tHb] versus RR ($p_{\mathrm{FDR}}=0.029$, $p=0.016$), and [tHb] versus log(PRQ) ($p_{\mathrm{FDR}}<0.0001$, $p<0.0001$).

Fig. 3

(a) Correlation matrix of bivariate scatter plots of $\mathrm{\Delta }{\mathrm{StO}}_2$, $\mathrm{\Delta }\mathrm{tHb}$, ${\mathrm{StO}}_2$, [tHb], ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$, HR, RR, and log(PRQ). The best nonlinear fit is presented for pairs with a significant correlation (red: significant correlation proved by both FDR-corrected and uncorrected $p$ values; blue: only uncorrected $p$ value). The level of significance is calculated from goodness-of-fit results. The red and blue shaded areas show 95% of confidence intervals. Outliers are not displayed. (b) Bar chart of $r$- and $\epsilon$ values for each pair of $\mathrm{\Delta }{\mathrm{StO}}_2$, $\mathrm{\Delta }\mathrm{tHb}$, ${\mathrm{StO}}_2$, [tHb], ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$, HR, RR, and log(PRQ) parameters (sorted in descending order of $r$-values). $\epsilon$ values $<-0.5$, near-zero ($-0.5<\epsilon <0.5$), between 0.5 and 1, and $>1$ are indicative of no, inconclusive, moderate, and strong correlation, respectively. (c) Bar chart of ICC values for all cerebral and cardiorespiratory parameters. Error bars represent the 95% confidence interval. The reliability of the ICC is indicated.

Correlations were also observed between the cerebral parameters: $\mathrm{\Delta }{\mathrm{StO}}_2$ versus $\mathrm{\Delta }\mathrm{tHb}$ ($p_{\mathrm{FDR}}=0.008$, $p=0.003$), $\mathrm{\Delta }{\mathrm{StO}}_2$ versus ${\mathrm{StO}}_2$ ($p_{\mathrm{FDR}}<0.0001$, $p<0.0001$), $\mathrm{\Delta }{\mathrm{StO}}_2$ versus [tHb] ($p_{\mathrm{FDR}}=0.020$, $p=0.008$), and ${\mathrm{StO}}_2$ versus [tHb] ($p_{\mathrm{FDR}}<0.0001$, $p<0.0001$). A relatively strong correlation between cardiorespiratory parameters was also present: ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$ versus HR ($p_{\mathrm{FDR}}<0.0001$, $p<0.0001$), ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$ versus RR ($p_{\mathrm{FDR}}=0.039$, $p=0.022$), ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$ versus log(PRQ) ($p_{\mathrm{FDR}}<0.0001$, $p<0.0001$), HR versus RR ($p_{\mathrm{FDR}}=0.001$, $p<0.0005$), HR versus log(PRQ) ($p_{\mathrm{FDR}}<0.0001$, $p<0.0001$), and RR versus log(PRQ) ($p_{\mathrm{FDR}}<0.0001$, $p<0.0001$). Since PRQ was calculated from HR and RR, we expected a linear correlation of HR versus log(PRQ), and RR versus log(PRQ). Additionally, Fig. 3(b) illustrates the coefficient of correlation “$r$” and BSE parameter “$\epsilon$” for each pair. The order of correlation starting from strongest ($r$-value close to 1 and $\epsilon >0.5$) is as follows: (1) ${\mathrm{StO}}_2$ versus ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$, (2) ${\mathrm{StO}}_2$ versus [tHb], (3) ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$ versus HR, (4) [tHb] versus log(PRQ), (5) ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$ versus log(PRQ), (6) $\mathrm{\Delta }{\mathrm{StO}}_2$ versus ${\mathrm{StO}}_2$, (7) HR versus RR, and (8) [tHb] versus HR.

3.4.

Intrasubject Variability: ICC Values Indicate Good-to-Excellent Reliability of Most Parameters

Figure 3(c) presents the ICC values of all variables. The ICC of ${\mathrm{StO}}_2$, [tHb], ${\mathrm{StO}}_2$ (left), [tHb] (right), [tHb] (left), ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$, HR, and RR indicates excellent reliability. The ICC of the PRQ, ${\mathrm{StO}}_2$ (right), and $\mathrm{\Delta }\mathrm{tHb}$ shows good and the ICC of $\mathrm{\Delta }{\mathrm{StO}}_2$ represents moderate reliability.

3.5.

Dependence of Cerebral Parameters on Sex and Seasonal Changes

The impact of sex and seasons (infradian changes) on cerebral parameters is depicted in Fig. 4. A significant difference due to sex ($p<0.0001$) was observed in both ${\mathrm{StO}}_2$ and [tHb]. In addition, $\mathrm{\Delta }[\mathrm{tHb}]$ was higher for males compared to females ($p<0.001$), but not for $\mathrm{\Delta }{\mathrm{StO}}_2$ ($p=0.635$). We found that FCOA showed the same trend of seasonal changes for both female and male groups and was higher in autumn and winter compared to spring and summer (spring versus autumn, $p=0.001$; spring versus winter, $p<0.001$; summer versus autumn, $p=0.024$; summer versus winter, $p<0.001$ and autumn versus winter, $p=0.011$). Interestingly, a Cosinor model fitted to [tHb] data represents completely contrary patterns for male and female groups. Absolute [tHb] values of males were higher in spring and summer than that of in autumn and winter. Conversely, the [tHb] values of females were observed at higher levels in autumn and winter in comparison with spring and summer.

Fig. 4

Changes in ${\mathrm{StO}}_2$, [tHb], $\mathrm{\Delta }{\mathrm{StO}}_2$, and $\mathrm{\Delta }\mathrm{tHb}$ due to time of year [infradian changes (a)–(d)] and time of day [circadian changes (e)–(h); morning: 7:00–9:30 and 9:30–12:00; afternoon: 12:00–14:30 and 14:30–17:00; evening: 17:00–19:30 and 19:30–22:00] for females (orange) and males (green). Cosinor model and the sum of 2 cosines function are fitted to infradian and circadian changes, respectively (female, dark orange lines; male, dashed green lines). Error bars represent the 95% confidence interval.

3.6.

Dependence of Cerebral Parameters on Time of Day

The effect of time of day (circadian changes) on cerebral parameters is also shown in Fig. 4. A sum of 2 cosine functions was applied to fit the circadian rhythm of the data. The same trend of higher ${\mathrm{StO}}_2$ values in the morning was observed for males and females ($p<0.001$). Males demonstrated very high [tHb] values in the early morning and late evening ($p<0.001$), but the trend was almost opposite in females ($p<0.05$). The highest FCOA for [tHb] values were found at 10:00 and 14:00 for males and females, respectively. Regardless of sex, a highly significant difference was found between ${\mathrm{StO}}_2$ values of morning and afternoon ($p<0.001$). There was also a significant difference between ${\mathrm{StO}}_2$ values of morning and evening ($p=0.011$). No significant changes were found for [tHb] (morning versus afternoon: $p=0.061$; morning versus evening: $p=0.17$).

3.7.

Dependence of Cerebral Parameters on Temperature

Figure 5 shows the changes in room temperature with respect to season 5(a) and time of day 5(b), and the temperature dependency of cerebral parameters [5(c)–5(f)]. The mean room temperature was $22.8°\mathrm{C}\pm 0.6°\mathrm{C}$ (range: 20.8°C to 24.8°C). Since the room was not air-conditioned, time of day and seasonal changes had an impact on the room temperature. As expected, the maximum room temperatures were recorded in summer and during the late evening. There was no significant linear correlation between room temperature and cerebral. The only exception was $\mathrm{\Delta }{\mathrm{StO}}_2$, which decreased with increasing room temperature (females: $r=-0.35$, $p<0.0001$; males: $r=-0.30$, $p<0.0002$).

Fig. 5

Changes in temperature with (a) time of year and (b) time of day. Dependence of (c) ${\mathrm{StO}}_2$, (d) [tHb], (e) $\mathrm{\Delta }{\mathrm{StO}}_2$, and (f) $\mathrm{\Delta }\mathrm{tHb}$ on temperature. The lines represent a linear fit and the shaded areas and error bars show 95% confidence intervals. Outliers are not displayed.

3.8.

Dependence of Cerebral Parameters on Mood and Chronotype

The mean valence, arousal, and dominance ratings during resting state assessed by SAM scales (ranging from 1 to 5) was $4.02\pm 0.69$, $2.92\pm 0.91$, and $3.24\pm 0.77$, respectively. The dependence of cerebral parameters on mood and chronotype is displayed in Fig. 6. No correlation was observed between the cerebral parameters and the positive affect scale of the PANAS questionnaire. The Horne and Östberg index was calculated for each subject to measure the chronotype. The highest numbers (score: 59 to 86) indicate morningness and the lowest numbers (score: 16 to 41) eveningness. Scores from 42 to 58 indicate neither morningness nor eveningness. We found a linear correlation between ${\mathrm{StO}}_2$ versus chronotype ($r=-0.1$, $p=0.03$), and [tHb] versus chronotype ($r=-0.09$, $p=0.04$).

Fig. 6

The scatter plots show the dependence of ${\mathrm{StO}}_2$, [tHb], $\mathrm{\Delta }{\mathrm{StO}}_2$, and $\mathrm{\Delta }\mathrm{tHb}$ on chronotype (Horne-Östberg index) and mood (positive affect scale of PANAS). Correlations between the data are indicated by a linear fit, and the red shaded areas indicate 95% confidence intervals. Outliers are not displayed.

4.1.

Absolute Values of Tissue Oxygenation and Hemoglobin Concentration

The normal range of ${\mathrm{StO}}_2$ and [tHb] of the brain was investigated for medical applications. These absolute values are approximately in accordance with the literature.^31,32 The absolute ${\mathrm{StO}}_2$ values of the PFC are in good agreement with Choi et al. (R-PFC: $74.75\%\pm 5.83\%$ versus. L-PFC: $75.63\%\pm 5.86\%$; $N=30$, age: 20 to 50 years, device: Imagent, ISS Inc.).³³ However, our absolute [tHb] values at the PFC are lower in comparison with their findings (R-PFC: $79.68\pm 12.15\mu \mathrm{M}$, L-PFC: $76.93\pm 14.98\mu \mathrm{M}$). Our [tHb] values are similar to those of Vernieri et al. (left versus right frontal region: $46.2\pm 11.9\mu \mathrm{M}$ versus $44.0\pm 12.9\mu \mathrm{M}$; 30 subjects, age: $63.9\pm 8.2$, device: Oximeter, ISS Inc.).³⁴ Moreover, our ${\mathrm{StO}}_2$ values are a bit higher compared to our previous study [${\mathrm{StO}}_2$ (right) = 68.6% (IQR: 63.5% to 72.4%), ${\mathrm{StO}}_2$ (left) = 56.8% (IQR: 52.9% to 63.4%); 24 subjects, age: $22.0\pm 6.4$ years, device: OxiplexTS, ISS Inc.].¹ The reasons for the difference between studies may be the age, physiological state of subjects, or methodology.

4.2.

Absolute Values of Cardiorespiratory Parameters

Arterial partial pressure of carbon dioxide (${\mathrm{PaCO}}_2$) is one of the strongest parameters that affect CBF and [tHb].¹⁹ Therefore, ${\mathrm{PaCO}}_2$ has been included in functional brain studies to ensure a correct interpretation of the signals.^35–37 We measured ${\mathrm{PaCO}}_2$ by the ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$ method, which also provides continuous and noninvasive RR. Our ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$ values were in agreement with the literature.^38–41

Our findings showed that the mean value of HR was $68\pm 11$ BPM (range: 41 to 116 BPM, females: $69\pm 11$ BPM, males: $66\pm 11$ BPM; $p=0.036$). In 35,000 healthy subjects, a mean HR of 72 BPM (age: 20 and over, females: $74\pm 0.2$ BPM, males: $71\pm 0.3$ BPM; $p<0.05$) was determined, which is close to our results.⁴²

The mean RR value measured in our study was $16.5\pm 2.9$ BrPM, which is within typical RR for adults (range: 6.9 to 27.1 BrPM, females: $16.8\pm 2.8$ BPM, males: $16.1\pm 3.1$ BPM; $p=0.015$).

The PRQ is a parameter to attain the overall current state of human physiology.⁴³ PRQ represents the state of the ANS and is a measure of cardiorespiratory coordination. PRQ is time- and sex-dependent, and changes during human development, physical activity, and body posture with specific patterns during sleep.⁴³ The resting state PRQ distribution has a peak at $\sim 4$.^44–46 We also found a mean resting state PRQ of $\sim 4$ ($4.2\pm 0.9$, ranging from 2.1 to 9.0).

4.3.

Relationships with Systemic Physiology

Although systemic physiological activity affects the absolute values of ${\mathrm{StO}}_2$ and [tHb], FCOA was not influenced. The reason is that both R-PFC and L-PFC are affected in the same way by systemic physiology.^1,21 Although we generally confirmed this finding, we found nonlinear correlations between $\mathrm{\Delta }{\mathrm{StO}}_2$ versus RR ($p=0.022$) and $\mathrm{\Delta }{\mathrm{StO}}_2$ versus PRQ ($p=0.024$).

As expected, we found a highly significant ($R^2=0.10$, $p<0.0001$) positive linear correlation between ${\mathrm{StO}}_2$ and ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$. Our finding is in agreement with the study carried out by Miller and Mitra.⁴⁷ and with the physiologically well-known ${\mathrm{CO}}_2$-response, i.e., a decrease in ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$ (hypocapnia) reduces the CBF by cerebral vasoconstriction.⁴⁸ This reduced oxygen supply leads to a lower ${\mathrm{StO}}_2$.^35,36 Hence, ${\mathrm{P}}_{\mathrm{ET}}{\mathrm{CO}}_2$ is positively correlated with ${\mathrm{StO}}_2$

4.4.

Frontal Cortex Oxygenation Asymmetry

In EEG studies, activity in left frontal regions is mostly associated with appetitive motivation and approach-related affect such as hope, happiness, and joy (positive affect). Conversely, the right frontal regions are related to vigilant attention and behavioral inhibition that regularly occurs during certain withdrawal-related affect such as depression and nervousness (negative affect).^{11,12,49–51} In general, the right frontal cortex reflects motivational systems of approach and avoidance, whereas the left frontal cortex inhibits the amygdala and downregulates negative affect.^52,53 Higher right frontal activity is attributed to greater negative affect (e.g., film-induced fear and disgust), whereas positive affect (e.g., film-induced happiness) elicits a higher left frontal activity.^54,55

We hypothesized that the FCOA reveals asymmetry of the PFC neuronal activity at rest. The ${\mathrm{StO}}_2$ was higher at the R-PFC than the L-PFC, indicating that the R-PFC is more activated than the L-PFC and this indicates a higher inhibitory activity or withdrawal motivation. This is reasonable considering that the subjects were in the resting state. Such rightward lateralization has also been found in the literature.^56,57 Although our findings are in line with several studies indicating that the right cortices have a stronger response compared to the left ones, some studies have reported no hemispheric differences or even leftward regional lateralization.^33,58 Liu et al.⁶ demonstrated that right or left regional laterality could be observed across different brain systems depending on multiple genetic or environmental mechanisms.