1. Introduction

Resting-state functional connectivity (RSFC), uncovered for the first time by Biswal, et al. in a functional magnetic resonance imaging (fMRI) study [1], refers to as the temporal synchrony in the low-frequency (e.g., <0.1 Hz) spontaneous activity in functionally related brain areas. It basically reflects the organization of brain functional networks, thus attracting much attention from a variety of scientific societies including fundamental neuroscience, neuroimaging and clinical applications [2]. To date, numerous RSFC studies have demonstrated altered RSFC in various neurological disorders, such as Parkinson's disease [3], dementia [2,4], Alzheimer’s disease [2–4], schizophrenia [2,4], depression [4], attention-deficit/hyperactivity disorder [4], and autism [5,6]. Therefore, the alterations in RSFC may have the potential to serve as biomarkers for neurological diseases.

Apart from fMRI, functional near-infrared spectroscopy (fNIRS) has also been widely used in RSFC studies [7–12]. Similar to fMRI measuring cerebral blood oxygenation signal (i.e., blood-oxygenation-level-dependent, BOLD), fNIRS measures the concentration change in oxygenated hemoglobin (HbO₂) and deoxygenated hemoglobin (Hb) in the cortex. Both BOLD and HbO₂/Hb are associated with neuronal activity. It has been demonstrated that BOLD and HbO₂/Hb are correlated with each other [13], reflecting the same neurophysiological origin but from different aspects.

As an optical brain imaging technique, fNIRS has several advantages, for example, it is safe, low-cost, easy-to-operate, and more tolerant to head movement as compared to other imaging modalities such as fMRI and Positron emission tomography (PET). In addition, it does not need special lab conditions required for performing measurements. These merits may render fNIRS a suitable tool for investigating brains of individuals such as children, in particular young children who may not be able to keep their heads motionless during the measurement. Less requirement to the lab condition makes it possible for using fNIRS to perform screening for neurological disorders even during a clinic visit.

Due to the neurovascular coupling, neural activity is associated with alternations in local cerebral hemodynamic variables, including blood oxygenation, blood volume, blood flow and metabolic rate of blood oxygenation. Each of these variables can explain what is going on in the cortex in terms of its own respect, thus the more we know about the hemodynamic variables, the more we may know about the brain.

As a relatively new optical brain imaging technique, near-infrared diffuse correlation spectroscopy (DCS) is very similar to fNIRS in appearance (in fact, where fNIRS can perform measurements, DCS can), but measures the relative cerebral blood flow (i.e., blood flow index, BFi) instead of blood oxygenation [14–20]. It has been used for studying brain functions [21–24] and bedside monitoring of brain in various clinical situations such as ischemic stroke and traumatic brain injury [14,25]. Despite the similarity in the appearance of system, the principle behind fNIRS and DCS is quite different. fNIRS measures the optical density (OD) of the emitted light and then converts OD into the hemoglobin concentration, while DCS records the temporal fluctuation of the emitted light (i.e., temporal intensity autocorrelation function g₂(τ), from which the blood flow information can be extracted). Therefore, different from BOLD measured by fMRI and hemoglobin concentration measured by fNIRS, DCS provides a measure of cerebral blood flow (CBF) which is also closely associated with the brain activity.

In a recent RSFC study, DCS was used to record the prefrontal spontaneous fluctuations, and for the first time the CBF-RSFC was demonstrated [26]. However, in this pilot study of using DCS to detect RSFC, only intra-hemispheric RSFC was measured, e.g., CBF-RSFC within the left hemisphere and the right hemisphere. No inter-hemispheric RSFC was detected and investigated in terms of CBF. Since significant alterations in the inter-hemispheric RSFC have been observed in a number of neurological disorders such as autism [5,6,27,28] and affective disorders [29], the inter-hemispheric RSFC has the potential to serve as a biomarker for diagnosing neural diseases. For example, based on the alteration in the inter-hemispheric BOLD-RSFC, the prediction on autism can be achieved with a sensitivity of 72.4%, and a specificity of 83.7% [6]. Recent fNIRS studies have also demonstrated the reduced inter-hemispheric HbO₂ /Hb-RSFC in autism [27,28], which has been attempted to make prediction on autism [28]. Therefore if inter-hemispheric RSFC could also be evaluated with CBF, the CBF-RSFC might become a new measure for studying brain functional connectivity and for predicting autism.

More hemodynamic variables could provide more information on the brain resting-state and RSFC, therefore understanding RSFC in terms of multi-hemodynamic variables, especially those variables with less or weak correlation between each other (which means they do not overlap with each other in providing information on hemodynamics), could be helpful for better understanding of brain networks, and beneficial to accurate prediction of neurological disorders. In a study on brain function with DCS, it has been demonstrated that the temporal responses of CBF and CBV are different to visual stimuli [24]. This discrepancy suggests the temporal variations of these two hemodynamics variables may show different behavior in the neurovascular coupling in processing the functional task. However, it’s still unclear on the relationship between the temporal variation of CBF and CBV in resting-state. If resting-state CBF and CBV do not vary in a same temporal way, the CBF- and CBV-RSFC could be different. Therefore, the inter-hemispheric CBF-RSFC would become a new measure independent of CBV-RSFC for evaluating the functional connectivity between the two hemispheres. To address these issues, in this study we used DCS to explore the inter-hemispheric RSFC in terms of CBF and CBV, and to investigate the relationship between CBF and CBV in resting-state. We believed if we could measure the inter-hemispheric CBF- and CBV-RSFC, then HbO₂/Hb-RSFC and CBF/CBV-RSFC could be combined together in the future for the study of brain functional networks and diagnosis of neurological diseases such as autism, since uncorrelated variables could provide different discriminative information for the diagnosis, leading to improved accuracy for the prediction of diseases.

2. Methods

2.1 DCS theory and an 8-channel DCS system

Diffuse correlation spectroscopy (DCS, in literature it is also called diffusing-wave spectroscopy, DWS) is an optical technique used for probing dynamics in turbid media such as latex, colloidal suspension [30,31], and biological tissues [19]. In DCS, laser light with a long coherence length is used to illuminate the medium. The emitted light from the surface consists of photons experienced many scattering events with moving scatterers. As long as the coherence length is longer than the typical path length of these photons, speckle pattern is formed on the surface due to the interference of the emitted light. Moving scatterers lead to fast temporal fluctuation of the speckles. DCS measures the temporal autocorrelation function of the light intensity (g₂ (τ) = < (I(t)*I(t + τ)) > / < I(t) >²) of a speckle to characterize the dynamics in the medium, for example, faster decay of g₂ (τ) means faster dynamics in the medium. In contrast to fNIRS measuring cerebral blood oxygenation, DCS detects the relative cerebral blood flow.

A custom-made 8-channel DCS system was used in this study, as shown in Fig. 1. The system consists of two lasers (DL785-150-SO, Crystalaser, USA) emitting 785 nm laser light with a long coherence length (>1 m). One laser is used for illuminating the left and the other for the right hemisphere. Each illumination laser beam is coupled into a multimode fiber (1000 µm core diameter, numerical aperture NA = 0.39) guiding light to the scalp. The emitted light from the scalp at each hemisphere is collected by 4 single-mode detection fibers (4.2 µm core diameter, numerical aperture NA = 0.14). The distance between the source fiber and detection fiber on the scalp is fixed to be 2.71 cm. Two 4-channel single photon counting modules (SPCM-AQ4C, Excelitas Technologies, Canada) are used to record photons. An 8-channel correlator (correlator.com, USA) is connected to the output ends of the two single photon counting modules, recording the light intensity and calculating the temporal intensity autocorrelation functions g₂ (τ). An USB cable connecting the correlator to a computer transmits the light intensity and time-resolved g₂ (τ) to the computer serving as data display and storage. A home-made rubber cushion is used for fixing the source and detector fibers. To ensure a good fiber-scalp optical coupling and comfortable contact between fibers and scalp, we made a fixture for each fiber with an elastic contact between the fiber end and the scalp. The output of this system includes time-resolved temporal intensity autocorrelation function g₂ (τ) and light intensity for 8 channels.

 

Fig. 1. The sketch of the 8-channel DCS system for detecting brain function.

Download Full Size | PDF

2.2 Experimental protocol

Twenty-one young (23.7 ± 1.2 years old, 8 males and 13 females) healthy subjects recruited from students in the South China Normal University participated in the experiments. They were all right-handed and had no history of any neurological disease. During the DCS measurement, 10-min spontaneous activity was recorded from the bilateral prefrontal cortex on each subject. The integration time for g₂ (τ) was 2 seconds, while the temporal resolution for the light intensity was 0.105 second. On each hemisphere 4 channels (each channel consists of a pair of source and detector) were positioned over the dorsolateral prefrontal cortex (DLPFC), as shown in Fig. 2. The illumination power at the output ends of the two source fibers were kept consistent at 75 mw. To ensure the power density on the scalp was below the safety limit (e.g., 4 mw/mm²), the fiber ends did not contact directly to the skin, but were kept a distance of 6 mm from the skin surface. The international 10/20 system was referenced for locating the optodes (source or detection fibers). The measurements were performed in a dark and quiet room. During the measurement, subjects seated in a comfortable chair, closed their eyes. They were instructed to do nothing, avoid falling asleep, remain quiet and keep body as motionless as possible. Before the measurement, all subjects were informed about the experimental protocol, and a written consent was signed by each of them. The experimental protocol was approved by the Ethic Committee of South China Normal University.

 

Fig. 2. Optode locations for measuring the spontaneous activity with the DCS. 4 channels located on the left and the other 4 channels on the right dorsolateral prefrontal cortex (DLPFC). One source was fixed at AF7 (left) and the other source at AF8 (right). D1-D8 indicated 1-8 detectors. The international 10/20 system was referenced for locating the optodes. The source-detector spacing was 2.71 cm.

Download Full Size | PDF

2.3 Data analysis

The main purpose of this study was to estimate the inter-hemispheric CBF- and CBV-RSFC, and investigate the relationship between CBF and CBV in resting-state. To extract the cerebral BFi, a widely used approach is to fit the g₂ (τ) curve with a semi-infinite tissue model in which the Brownian motion is adopted to describe the dynamics (i.e. CBF) [17,21,26]. However, in this approach to accurately extract the BFi from g₂ (τ), the optical parameters such as the absorption and reduced scattering coefficient need to be known before the fit [32]. These optical parameters could be measured separately by another optical setup such as time-domain or frequency-domain NIRS, or taken from published literature.

Since in this RSFC study we only need relative CBF for calculating the Pearson correlation coefficient, we may simplify the data processing by simply using the decay rate of g₂ (τ)-1 instead of using the BFi (the Brownian particle diffusion coefficient D_B times a factor α, i.e., αD_B [17]). To demonstrate this, we performed a simulation study. In the simulation, we assumed to know a time series of BFi, and generated time-resolved g₂ (τ) using the semi-infinite tissue model, under conditions of various optical properties: the absorption coefficient µ_a= (0.10-0.45) cm^-1, and the reduced scattering coefficient µ_s^’= (5-15) cm^-1, sufficiently covering the range of tissue optical parameters. Once the g₂ (τ) was obtained, we used a single exponential decay function, y(τ)=β*exp(-τ/τ₀) to fit the g₂ (τ) -1 curve, and obtained the decay rate R=1 / τ₀. We then calculated the correlation coefficient between the time-series of R and the known BFi, for many combinations of µ_a and µ_s^’, and showed the result in Fig. 3. In a typical range of optical parameters in human head tissue, e.g., µ_a ∼ 0.19 cm^-1, µ_s^’ ∼ (6-11) cm^-1 [33], the correlation values between the BFi and decay rate R were larger than 0.98, demonstrating that when calculating RSFC, using the decay rate R instead of using BFi is acceptable and rather accurate.

 

Fig. 3. Correlation coefficient between the BFi and decay rate R of g₂ (τ)-1. In a wide range of combinations of µ_a (0.10-0.45 cm^-1) and µ_s’ (5-15 cm^-1), the correlation coefficient was larger than 0.92, indicating the two variables were highly correlated.

Download Full Size | PDF

Before calculating CBF-RSFC, several data processing steps were taken: (1) use a 2nd order polynomial fit to get ride of the 1st and 2nd order drift in the time-series of R; (2) a zero-phase 2nd order Butterworth filter with a pass-band of 0.009 to 0.08 Hz was applied [9,12]; (3) a independent component analysis (ICA) based algorithm was adopted to suppress the global component and motion artifact [28,34]; (4) compute the Pearson correlation coefficient for each detector pair locating symmetrically on the two hemispheres, e.g., pair (1,5) consisting of D1 and D5 (see Fig. 2).

For calculating the CBV-RSFC, we first estimated the relative CBV based on the recorded light intensity. We know the change in the optical density (OD) is proportional to the change in the absorption coefficient (ΔOD∝Δµ_a). We assumed in resting-state we could use ΔOD instead of CBV or HbT (=HbO₂+Hb) to compute CBV-RSFC (see Appendix). Thus in calculating CBV-RSFC, we calculated correlation coefficients for all symmetric detector pairs using the variable ΔOD. The other steps were same as those used for calculating CBF-RSFC.

In order to investigate the relationship between CBF and CBV, the Pearson correlation between these two hemodynamic variables were calculated for each detector on all the subjects.

Since in 21 subjects there were 8 males and 13 females, the sex-related differences in CBF- and CBV-RSFC were also presented.

To test the statistical significance in the correlation coefficients, Fisher-Z transformation was used to transfer the correlation coefficient value r to Z value, and then T-test was performed. The significance level was set to be 0.05. For multiple comparisons, the false discovery rate (FDR) correction was used.

3. Result

There were 4 detectors locating on the left and the other 4 detectors on the right DLPFC, thus forming 4 detector pairs crossing between the two hemispheres. Each pair consisted of two detectors locating symmetrically on the left and right side. The correlation coefficient for each pair indicated the inter-hemispheric RSFC, as shown in Fig. 4.

For the inter-hemispheric CBF-RSFC on the DLPFC, the correlation coefficient is 0.58 ± 0.17, 0.49 ± 0.17, 0.66 ± 0.12, 0.66 ± 0.12 for detector pair (1, 5), (2, 6), (3, 7) and (4, 8), respectively. For the inter-hemispheric CBV-RSFC on the DLPFC, the correlation coefficient is 0.72 ± 0.12, 0.68 ± 0.11, 0.66 ± 0.16, 0.77 ± 0.09 for detector pair (1, 5), (2, 6), (3, 7) and (4, 8), respectively.

T-test shows that all these correlation coefficients are significantly larger than zero, demonstrating the inter-hemispheric RSFC on the DLPFC in terms of CBF and CBV. In addition, T-test further reveals that for the 4 correlation coefficients on CBF and CBV, there is no significant difference between each other.

 

Fig. 4. Inter-hemispheric CBF-RSFC (a) and CBV-RSFC (b) for the DLPFC. The x-axis indicates the detector pairs, e.g., pair (1,5) means the pair consisting of detector 1 (locating on the left side) and detector 5 (locating symmetrically on the right side). The y-axis is the correlation coefficient. The error bars are the standard deviations calculated from the all 21 subjects.

Download Full Size | PDF

Taking the whole measured area at each hemisphere (i.e., the region covered by the 4 detectors at each hemisphere) as a region-of-interest (ROI), we could average the 4 correlation coefficients to obtain the CBF- and CBV-RSFC between the two ROIs for each subject, and then calculate the grand means for the all 21 subjects. We obtained CBF-RSFC=0.60 ± 0.12 and CBV-RSFC=0.71 ± 0.09. T-test showed the CBV-RSFC was significantly stronger than CBF-RSFC (p = 0.0023).

Considering sex as a variable, we could calculate CBF- and CBV-RSFC between the left and right ROIs for each sex, and analyze sex-related differences between the two groups. For the male group, CBF-RSFC = 0.57 ± 0.11, CBV-RSFC=0.76 ± 0.07. For the female group, CBF-RSFC = 0.62 ± 0.12, CBV-RSFC=0.68 ± 0.08. The T-test showed the sex-related difference was not significant in CBF-RSFC, but marginally significant in CBV-RSFC (p = 0.049).

The group-averaged correlation coefficient between CBF and CBV for each of the 8 detectors is shown in Fig. 5. T-test shows there is no significant difference between each other. We averaged the correlation coefficients over the 8 detectors for each subject to get the mean correlation coefficient, and then averaged the mean correlation coefficients over the 21 subjects to obtain the grand mean of the correlation coefficients between CBF and CBV. The grand mean was 0.47 ± 0.11, indicating the two hemodynamic variables showed moderate correlation in the measured prefrontal area.

 

Fig. 5. Correlation coefficients between time series of CBF and and CBV for the 8 detectors. The error bars are standard deviations calculated from the all 21 subjects.

Download Full Size | PDF

4. Discussion

Cerebral blood hemodynamic variables include hemoglobin (HbO₂/Hb) concentration, cerebral blood flow (CBF), and cerebral blood volume (CBV). The fMRI BOLD signal is a measure of cerebral blood oxygenation level closely related to the hemoglobin concentration, e.g., (1/BOLD) is negatively correlated to HbO₂, and positively correlated to Hb when brain is functionally activated [35]. The cerebral metabolic rate of oxygen (CMRO₂) is also an index for measuring the activation of brain. However, the CMRO₂ can be derived from HbO₂ / Hb and CBF [21]. The CBV is proportional to the total hemoglobin concentration (HbT), which is a sum of HbO₂ and Hb (i.e., HbT = HbO₂+Hb). Therefore, among all these hemodynamic variables, the cerebral blood hemoglobin concentration (HbO₂/Hb) and CBF are more fundamental variables. fNIRS measures the cerebral blood hemoglobin (HbO₂/Hb/HbT), while DCS measures the CBF and CBV, providing a different measurement to fNIRS. The combination of these two optical techniques are able to provide more comprehensive information on the cortical hemodynamics, which could be helpful for better understanding of healthy brains and brains with neurological diseases. For example, in the previous studies with fNIRS [27,28], we and our colleagues have already observed significant reduction in inter-hemispheric HbO₂- and Hb-RSFC in autism. Including the CBF-RSFC as a new variable, we may obtain a more accurate prediction on autism.

In this study, we obtained the prefrontal inter-hemispheric CBF-RSFC=0.60 ± 0.12 and CBV-RSFC=0.71 ± 0.09. Since the prefrontal fNIRS-RSFC has been revealed in several fNIRS studies [9,29], we made comparisons between our findings and the previous reports with fNIRS on healthy adults. Mesquita, et al. measured resting-state brain activity from 11 healthy adult male subjects [9], and obtained the prefrontal RSFC: HbO₂-RSFC=0.50 ± 0.12, Hb-RSFC=0.56 ± 0.15, HbT-RSFC=0.62 ± 0.20. Wu, et al. collected prefrontal resting-state fNIRS data from 62 healthy adults (32 males and 30 females) [29], and obtained inter-hemispheric HbO₂-RSFC for three ROIs: the inferior frontal gyrus (IFG), middle frontal gyrus (MFG) and superior frontal gyrus (SFG). The HbO₂-RSFC was 0.74 ± 0.12 for IFG, 0.73 ± 0.12 for MFG and 0.85 ± 0.08 for SFG. The ROI of the MFG in Wu, et al’s. study is closer to where we measured in the present study. The strengths of CBF-RSFC and CBV-RSFC we obtained are within the range of these two fNIRS studies.

In our previous study [26] using DCS to measure CBF-RSFC, we obtained the intra-regional (within DLPFC) CBF-RSFC=0.64 ± 0.25, and inter-regional (between DLPFC and the inferior prefrontal cortex) CBF-RSFC=0.34 ± 0.27 in the left hemisphere; and intra-regional CBF-RSFC=0.62 ± 0.23, and inter-regional CBF-RSFC=0.34 ± 0.26 in the right hemisphere. The inter-hemispheric CBF-RSFC we obtained in this study is 0.60 ± 0.12, which is is similar in strength to the previously identified intra-regional CBF-RSFC, but larger than the inter-regional CBF-RSFC. This is line with the fact that the connectivity between the functionally related areas (even locating apart, e.g., distributed over the two hemispheres) is stronger than the connectivity between functionally different areas locating even side by side in the same hemisphere.

The present study demonstrated both inter-hemispheric CBF- and CBV-RSFC on the DLPFC. However, the strength of CBF-RSFC is different from the CBV-RSFC for any of the 4 pairs. To elucidate this, we investigated the relationship between the two hemodynamic variables (CBF and CBV) by calculating the correlation coefficient between the CBF and CBV time series for each of the 8 detectors. Figure 5 shows that there is a moderate correlation (e.g, grand mean = 0.47 ± 0.11) between the temporal fluctuation of CBF and CBV in resting-state. This implies that the information provided by the two variables on the cerebral hemodynamics do not fully overlap with each other, instead they only partially overlap, indicating they could provide different knowledge on the mechanism of the cerebral hemodynamics in resting-state.

It is worth noting that in a resting-state fMRI study [36] performed on young (28 ± 5 years) health subjects, CBF and CBV were measured by using pseudo-continuous arterial spin labeling (pCASL) and gradient and spin echo (GRASE) based vascular space occupancy (VASO), respectively. The spontaneous fluctuations of CBF and CBV were observed to be different. The measured correlation maps in terms of CBF and CBV were also different. Therefore, our results are basically in line with these observations. The discrepancy in the spontaneous CBF and CBF was simply attributed to biological reasons, no detailed explanation was given in this fMRI study. In a recent fMRI modeling study on resting-state [37], a modified balloon model was used to describe the dynamic relationship among neuronal activity, BOLD, CBF, CBV and CMRO₂. In this model, there are two parameters describing the dynamic relationship between resting-state CBF and CBV, one is the exponent describing steady state flow–volume coupling, the other is characteristic time constant for CBV response. As long as the time constant is larger than zero, there will be a delay in CBV with respect to CBF. This time delay stands for delayed perfusion and will certainly cause reduction in the correlation coefficient between CBF and CBV. Therefore, according the modified balloon model [37], the delayed CBV response with respect to the change in CBF could be one of the reasons for explaining the discrepancy in the temporal fluctuations of CBF and CBV, and for explaining why the correlation coefficient between resting-state CBF and CBV is less than one.

To further explain the reason why CBF-RSFC is different from CBV-RSFC in our data, we show in Fig. 6 the difference between CBF- and CBV-RSFC with respect to the correlation coefficient (or coupling) between CBF and CBV (i.e., corr (CBF, CBV)) for each subject. The difference between CBF- and CBV-RSFC is negatively correlated (e.g., r=-0.44) with the correlation between CBF and CBV. This implies that when the correlation coefficient between CBF and CBV is larger, the difference between CBF- and CBV-RSFC is smaller. The relationship between (CBV-RSFC) - (CBF-RSFC) and Corr (CBF, CBV) might be easily understood in signal space. The correlation coefficient between two standardized signals (e.g., standardized S=(S-mean(S))/std(S)) is the cosine of the angle spanned by the two signals. Higher correlation between the two signals indicates the smaller angle between them. If the correlation between CBF and CBV is higher, the angle between CBF and CBV is smaller, which means the time series of CBF and CBV are closer or similar in signal space. Therefore, the difference between CBF- and CBV-RSFC must be smaller. Theoretically, if Corr (CBF, CBV) = 1, then CBF-RSFC = CBV-RSFC. However, in our measurements, the Corr (CBF, CBV) was always smaller than 1, leading to the difference between CBF- and CBV-RSFC. It has been reported that brain diseases such as cerebral ischaemia [38] and traumatic brain injury [39] may cause reduction in the correlation between CBF and CBV. This might be reflected by the alteration in CBF- and CBV-RSFC, and the difference between them.

 

Fig. 6. (CBV-RSFC) - (CBF-RSFC) with respect to the correlation between CBF and CBV (i.e., Corr (CBF, CBV)) for each of the 21 subjects. The straight line is a linear fit.

Download Full Size | PDF

The T-test shows the inter-hemispheric CBV-RSFC is significantly (p = 0.002) stronger than CBF-RSFC. This indicates in resting-state the temporal variation in CBV is more synchronous than CBF between the two DLPFCs. A possible explanation for this observation is that the change in CBF might be more sensitive to the local neuronal activity than CBV. Since if it is the case, the ‘small’ difference in neuronal activity between the two hemispheres may cause ‘big’ difference in the CBF between the two hemispheres, thus leading to smaller CBF- RSFC as compared to CBV-RSFC. This explanation may be in line with the observation that the change in CBF is larger than CBV in response to task-related brain activation, e.g., response to visual stimulation [40].

Sex-related differences in the inter-hemispheric CBF- and CBV-RSFC were also investigated. No significant difference was found in CBF-RSFC, whereas marginally significant difference was found in CBV-RSFC between the male and female group. The male group showed stronger functional connectivity in terms of CBV than the female group. This observation is similar to a previous fNIRS finding that the HbO₂-RSFC between the left and right middle frontal gyrus (MFG) is stronger in the males than females [35]. Since the number of subjects for each sex is small (8 males, 13 females) in the present study, the power of the statistical analysis may not be strong enough for analyzing the sex-related difference, we did not go further to explain the reason why the inter-hemispheric RSFC showed the sex-related difference in terms of CBV, but not in terms of CBF.

Significant alterations in both inter-hemispheric fMRI BOLD-RSFC [6] and fNIRS HbO₂/Hb-RSFC [27,28] have been observed in autism spectrum disorder (ASD), which has also been demonstrated having potential to serve as biomarkers for predicting ASD. More characteristic features associated ASD are revealed, more accurate prediction of ASD can be achieved. Including DCS for simultaneously measuring CBF and CBV, it is likely for us to observe alterations in CBF- and CBV-RSFC in ASD, which might become additional characteristic features associated with ASD, thus helpful for improving the accuracy in the prediction on ASD.

5. Conclusion

In summary, we used an 8-channel DCS system to measure the spontaneous hemodynamic fluctuations on the bilateral prefrontal cortex. By simulation, we have demonstrated when computing the correlation coefficient on CBF between a pair of detectors, it is not necessary to obtain BFi (or αD_B) based on a tissue model and known optical parameters. Instead, we may use a model-free parameter, the decay rate of the autocorrelation function g₂ (τ), to compute the correlation function on CBF. The inter-hemispheric RSFC on the prefrontal cortex was evaluated in terms of CBV and CBF. The two key hemodynamic variables, CBF and CBV, were moderately correlated in resting-state of brain, indicating each could provide different information on the mechanism of cerebral hemodynamics associated with the neuronal activity. Therefore, the combination of CBF- and CBV-RSFC could be helpful for better understanding of the functional networks in normal brains and brains with neurological disorders.

Appendix

The best wavelength for light to detect variation of CBV (or HBT) is around 800 nm, the isosbestic point where HbO₂ and Hb have the same extinction coefficient, since this wavelength is only sensitive to the change in HbT (HbT = HbO₂+Hb). In this study, we used light with 785 nm to detect CBV. To demonstrate this wavelength could be used for detecting CBV-RSFC, we did a study combining fNIRS measurements and simulation.

The resting-state spontaneous hemodynamic fluctuations were collected from the prefrontal cortex by our fNIRS setup (Foire-3000, Shimadzu Corp. Kyoto, Japan) [29]. From this fNIRS measurement, we obtained HbO₂, Hb and HbT time series. The same data analysis approach as that used in analyzing resting-state DCS decay rate R was applied to HbO₂, Hb and HbT time series. After we obtained the preprocessed time series, i.e., the low-frequency fluctuation in HbO₂, Hb, and HbT, we could calculate the temporal fluctuation of OD in 785 nm, i.e., ΔOD (785 nm), caused by the fluctuation of HbO₂ and Hb in resting-state. To do this, we used the extinction coefficient 0.7681 cm^-1nM^-1 for HbO₂ and 0.9975 cm^-1nM^-1 for Hb at 785 nm [41], and then we obtained ΔOD (785 nm) = 0.7681*Δ[HbO₂]+0.9975*Δ[Hb]. Figure 7 shows the comparison of HbT measured by fNIRS and the calculated ΔOD (785 nm) for a subject in the left prefrontal cortex. The correlation coefficient between HbT and ΔOD (785 nm) is 0.993.

 

Fig. 7. The measured HbT with fNIRS and calculated ΔOD (785 nm) using time series of HbO₂ and Hb measured from fNIRS in a subject (S1). The correlation coefficient between HbT and ΔOD (785 nm) is 0.993.

Download Full Size | PDF

Table 1 shows the correlation coefficients between HbT and the calculated ΔOD (785 nm) on 5 subjects. These results show very high correlation (>0.989) between HbT and the ΔOD (785 nm) in both left and right hemisphere for the 5 subjects. Therefore, it could be an acceptable assumption for using ΔOD (785 nm) instead of CBV (or HbT) to calculate CBV-RSFC (or CBV-related correlation coefficients) in resting-state.

Table 1. correlation coefficients between HbT and ΔOD (785 nm) in the left and right prefrontal cortex

View Table

Funding

National Natural Science Foundation of China (81771876); Guangdong Provincial Applied Science and Technology Research and Development Program (2017A010101023, 2017B030301007); Guangzhou Science and Technology Program key projects (2019050001); Special Funds for the Cultivation of Guangdong College Students' Scientific and Technological Innovation (pdjh2020b0155).

Disclosures

The authors declare that there are no conflicts of interest related to this article.

References

1. B. Biswal, F. Z. Yetkin, V. M. Haughton, and J. S. Hyde, “Functional connectivity in the motor cortex of resting human brain using echo-planar MRI,” Magn. Reson. Med. 34(4), 537–541 (1995). [CrossRef]  

2. M. P. Van Den Heuvel and H. E. H. Pol, “Exploring the brain network: a review on resting-state fMRI functional connectivity,” Eur. Neuropsychopharmacol. 20(8), 519–534 (2010). [CrossRef]  

3. C. Hohenfeld, C. J. Werner, and K. Reetz, “Resting-state connectivity in neurodegenerative disorders: Is there potential for an imaging biomarker?” NeuroImage-Clin. 18, 849–870 (2018). [CrossRef]  

4. M. Greicius, “Resting-state functional connectivity in neuropsychiatric disorders,” Curr. Opin. Neurol. 24(4), 424–430 (2008). [CrossRef]  

5. V. L. Cherkassky, R. K. Kana, T. A. Keller, and M. A. Just, “Functional connectivity in a baseline resting-state network in autism,” NeuroReport 17(16), 1687–1690 (2006). [CrossRef]  

6. I. Dinstein, K. Pierce, L. Eyler, S. Solso, R. Malach, M. Behrmann, and E. Courchesne, “Disrupted neural synchronization in toddlers with autism,” Neuron 70(6), 1218–1225 (2011). [CrossRef]  

7. B. R. White, A. Z. Snyder, A. L. Cohen, S. E. Petersen, M. E. Raich-le, B. L. Schlaggar, and J. P. Culver, “Resting-state functional connectivity in the human brain revealed with diffuse optical tomography,” NeuroImage 47(1), 148–156 (2009). [CrossRef]  

8. C. M. Lu, Y. J. Zhang, B. B. Biswal, Y. F. Zhang, D. L. Peng, and C. Z. Zhu, “Use of fNIRS to assess resting state functional connectivity,” J. Neurosci. Methods 186(2), 242–249 (2010). [CrossRef]  

9. R. C. Mesquita, M. A. Franceschini, and D. A. Boas, “Resting state functional connectivity of the whole head with near-infrared spectroscopy,” Biomed. Opt. Express 1(1), 324–336 (2010). [CrossRef]  

10. H. Zhang, L. Duan, Y. J. Zhang, C. M. Lu, H. Liu, and C. Z. Zhu, “Test-retest assessment of independent component analysis-derived resting-state functional connectivity based on functional near-infrared spectroscopy,” NeuroImage 55(2), 607–615 (2011). [CrossRef]  

11. L. Duan, Y. J. Zhang, and C. Z. Zhu, “Quantitative comparison of resting-state functional connectivity derived from fNIRS and fMRI: A simultaneous recording study,” NeuroImage 60(4), 2008–2018 (2012). [CrossRef]  

12. J. Li and L. N. Qiu, “Temporal correlation of spontaneous hemodynamic activity in language areas measured with functional near-infrared spectroscopy,” Biomed. Opt. Express 5(2), 587–595 (2014). [CrossRef]  

13. G. Strangman, J. P. Culver, J. H. Thompson, and D. A. Boas, “A quantitative comparison of simultaneous BOLD fMRI and NIRS recordings during functional brain activation,” NeuroImage 17(2), 719–731 (2002). [CrossRef]  

14. T. Durduran and A. G. Yodh, “Diffuse correlation spectroscopy for non-invasive, micro-vascular cerebral blood flow measurement,” NeuroImage 85, 51–63 (2014). [CrossRef]  

15. E. M. Buckley, A. B. Parthasarathy, P. E. Grant, A. G. Yodh, and M. A. Franceschini, “Diffuse correlation spectroscopy for measurement of cerebral blood flow: future prospects,” Neurophotonics 1(1), 011009 (2014). [CrossRef]  

16. D. A. Boas, C. Pitris, and N. Ramanujam, Handbook of Biomedical Optics, (CRC Press, 2011), Chap. 10.

17. C. Cheung, J. P. Culver, K. Takahashi, J. H. Greenberg, and A. G. Yodh, “In vivo cerebrovascular measurement combining diffuse near-infrared absorption and correlation spectroscopies,” Phys. Med. Biol. 46(8), 2053–2065 (2001). [CrossRef]  

18. R. C. Mesquita, T. Durduran, G. Yu, E. M. Buckley, M. N. Kim, C. Zhou, R. Choe, U. Sunar, and A. G. Yodh, “Direct measurement of tissue blood flow and metabolism with diffuse optics,” Philos. Trans. R. Soc., A 369(1955), 4390–4406 (2011). [CrossRef]  

19. D. A. Boas and A. G. Yodh, “Spatially varying dynamical properties of turbid media probed with diffusing temporal light correlation,” J. Opt. Soc. Am. A 14(1), 192–215 (1997). [CrossRef]  

20. D. A. Boas, S. Sakadžić, J. Selb, P. Farzam, M. A. Franceschini, and S. A. Carp, “Establishing the diffuse correlation spectroscopy signal relationship with blood flow,” Neurophotonics 3(3), 031412 (2016). [CrossRef]  

21. T. Durduran, G. Yu, M. G. Burnett, J. A. Detre, J. H. Greenberg, J. Wang, C. Zhou, and A. G. Yodh, “Diffuse optical measurement of blood flow, blood oxygenation, and metabolism in a brain during sensorimotor cortex activation,” Opt. Lett. 29(15), 1766–1768 (2004). [CrossRef]  

22. J. Li, G. Dietsche, D. Iftime, S. E. Skipetrov, G. Maret, T. Elbert, B. Rockstroh, and T. Gisler, “Noninvasive detection of functional brain activity with near-infrared diffusing-wave spectroscopy,” J. Biomed. Opt. 10(4), 044002 (2005). [CrossRef]  

23. F. Jaillon, J. Li, G. Dietsche, T. Elbert, and T. Gisler, “Activity of the human visual cortex measured non-invasively by diffusing-wave spectroscopy,” Opt. Express 15(11), 6643–6650 (2007). [CrossRef]  

24. J. Li, M. Ninck, L. Koban, T. Elbert, J. Kissler, and T. Gisler, “Transient functional blood flow change in the human brain measured noninvasively by diffusing-wave spectroscopy,” Opt. Lett. 33(19), 2233–2235 (2008). [CrossRef]  

25. L. He, W. B. Baker, D. Milej, V. C. Kavuri, R. C. Mesquita, D. R. Busch, K. Abramson, J. Y. Jiang, M. Diop, K. S. Lawrence, O. Amendolia, F. Quattrone, R. Balu, W. A. Kofke, and A. G. Yodh, “Noninvasive continuous optical monitoring of absolute cerebral blood flow in critically ill adults,” Neurophotonics 5(4), 045006 (2018). [CrossRef]  

26. J. Li, C. S. Poon, J. Kress, D. J. Rohrbach, and U. Sunar, “Resting-state functional connectivity measured by diffuse correlation spectroscopy,” J. Biophotonics 11(2), e201700165 (2018). [CrossRef]  

27. H. Zhu, Y. Fan, H. Guo, D. Huang, and S. He, “Reduced interhemispheric functional connectivity of children with autism spectrum disorder: evidence from functional near infrared spectroscopy studies,” Biomed. Opt. Express 5(4), 1262–1274 (2014). [CrossRef]  

28. J. Li, L. Qiu, L. Xu, E. V. Pedapati, C. A. Erickson, and U. Sunar, “Characterization of autism spectrum disorder with spontaneous hemodynamic activity,” Biomed. Opt. Express 7(10), 3871–3881 (2016). [CrossRef]  

29. S. Wu, L. Gao, C. Chen, J. Li, and S. He, “Resting-State Functional Connectivity in prefrontal cortex investigated by functional near-infrared spectroscopy: A longitudinal and cross-sectional study,” Neurosci. Lett. 683, 94–99 (2018). [CrossRef]  

30. G. Maret and P. E. Wolf, “Multiple light scattering from disordered media: The effect of Brownian motion of scatterers,” Z. Phys. B: Condens. Matter 65(4), 409–413 (1987). [CrossRef]  

31. D. J. Pine, D. A. Weitz, J. X. Zhu, and E. Herbolzheimer, “Diffusing-wave spectroscopy: dynamic light scattering in the multiple scattering limit,” J. Phys. 51(18), 2101–2127 (1990). [CrossRef]  

32. M. Diop, K. Verdecchia, T. Y. Lee, and K. St Lawrence, “Calibration of diffuse correlation spectroscopy with a time-resolved near-infrared technique to yield absolute cerebral blood flow measurements,” Biomed. Opt. Express 2(7), 2068–2081 (2011). [CrossRef]  

33. G. Strangman, M. A. Franceschini, and D. A. Boas, “Factors affecting the accuracy of near-infrared spectroscopy concentration calculations for focal changes in oxygenation parameters,” NeuroImage 18(4), 865–879 (2003). [CrossRef]  

34. J. Li and L. Qiu, “Temporal correlation of spontaneous hemodynamic activity in language areas measured with functional near-infrared spectroscopy,” Biomed. Opt. Express 5(2), 587–595 (2014). [CrossRef]  

35. G. Strangman, J. P. Culver, J. H. Thompson, and D. A. Boas, “A Quantitative Comparison of Simultaneous BOLD fMRI and NIRS Recordings during Functional Brain Activation,” NeuroImage 17(2), 719–731 (2002). [CrossRef]  

36. K. Zhang, D. Huang, and N. Jon Shah, “Comparison of Resting-State Brain Activation Detected by BOLD, Blood Volume and Blood Flow,” Front. Hum. Neurosci. 12, 443 (2018). [CrossRef]  

37. M. E. Archila-Meléndez, C. Sorg, and C. Preibisch, “Modelling the impact of neurovascular coupling impairments on BOLD-based functional connectivity at rest,” NeuroImage 218, 116871 (2020). [CrossRef]  

38. H. S. Markus, “Cerebral perfusion and stroke,” J. Neurol., Neurosurg. Psychiatry 75(3), 353–361 (2004). [CrossRef]  

39. G. J. Bouma and J. P. Muizelaar, “Cerebral blood flow, cerebral blood volume, and cerebrovascular reactivity after severe head injury,” J Neurotrauma 9(Suppl 1), S333–S348 (1992).

40. H. Ito, K. Takahashi, J. Hatazawa, S. G. Kim, and I. Kanno, “Changes in human regional cerebral blood flow and cerebral blood volume during visual stimulation measured by positron emission tomography,” J. Cereb. Blood Flow Metab. 21(5), 608–612 (2001). [CrossRef]  

41. S. J. Matcher, C. E. Elwell, C. E. Cooper, M. Cope, and D. T. Delpy, “Performance comparison of several published tissue near-infrared spectroscopy algorithms,” Anal. Biochem. 227(1), 54–68 (1995). [CrossRef]