2.1.

Subjects

Six healthy male volunteers [age ($\text{mean}\pm \mathrm{SD}$), $29.8\pm 4.7\text{years}$] participated in the present study. According to the Edinburgh Inventory,¹⁴ all were right-handed. Informed consent was obtained from each subject before the experiment. All investigations were in accordance with the latest version of the Declaration of Helsinki and were approved by the Institutional Review Board of the Niigata University School of Medicine.

2.2.

Motor Task

In a dimmed room, subjects were seated in a comfortable chair with their arms on the armrests. They faced the monitor, which presented a fixation cross at a distance of approximately 90 cm. A device, which counted the number of pinching movements and sent specific signals to the NIRS system, was placed at the end of the right armrest (Fig. 1). Subjects were instructed to make a simple right thumb-index finger pinch of this device every 10 s at their own pace. In order to minimize possible artifacts, they were asked to stare at the fixation cross on the screen and not count the number of pinching movements during the entire session. After 100 samples had been stored, the session was ended. The total testing time was about 17 min.

Fig. 1

Photograph of a custom-made trigger counter and two electrodes on the right first dorsal interosseus muscle used for electromyogram (EMG) recording. When a subject performs a simple pinch task, the number of pinching movements is detected by the counter, and an EMG input is triggered for near-infrared spectroscopy (NIRS) and the movement-related cortical potentials (MRCP) measurement system.

2.3.

MRCP Data Acquisition

EEG signals were recorded using $\mathrm{Ag}/\mathrm{AgCl}$ electrodes at Fz, Cz, C3, and C4, according to the International 10–20 system, with the left ear used as reference, and the forehead as ground [Fig. 2(a)]. EMGs associated with pinch movements were recorded with two electrodes placed on the first dorsal interosseus muscle of the right hand, and the signals for both the NIRS and the MRCP measurement system were triggered (Fig. 1). All data were digitized on-line at a rate of 500 Hz. The time constants for the EEG and EMG recordings were 3 and 0.03 s, respectively. The high-frequency filters for the EEG and EMG recordings were 70 and 1000 Hz, respectively. In all channels, a 50 Hz notch filter was applied.

Fig. 2

(a) Placement of the bilateral $4\times 4$ probe sets and electroencephalography (EEG) electrodes at Fz, Cz, C3, and C4 was in accordance with the International 10–20 system. The NIRS probes, which are separated from each other by 3 cm, are positioned over the bilateral motor cortex and motor association cortex. Fz, Cz, and C3 indicate the positions of EEG electrodes (the C4 electrode was also placed in the left hemisphere). (b) Schema of NIRS probes and EEG electrodes. The bilateral $4\times 4$ probe sets, positioned as channel numbers 18 and 42, are placed symmetrically beside Cz since C3 and C4 were enclosed by channel numbers 16, 19, 20, and 23 and numbers 40, 43, 44, and 47, respectively. The region of interest (ROI) is shaded gray: prefrontal cortex (PFC), dorsal premotor cortex (dPMC), and the hand area of the sensorimotor cortex (SMC). (c) Top view of 3-D brain cortex in subject 1, with the position of the probes on the brain (red dots: emitters; blue dots: detectors). PMC, dPMC, and the hand area of SMC were bilaterally covered by the placed probes. ROIs were set for PFC (channels 4, 5, 28, and 29), dPMC (channels 11, 12, 35, and 36), and the hand area of SMC (channels 16, 19, 20, 23, 40, 43, 44, and 47) as shown in (b).

2.4.

MRCP Data Analysis

For MRCP data analysis, the EEG recording was segmented into epochs from 1600 ms before EMG onset to 1500 ms after EMG onset. All movement trials were assessed by the interval time, the quantitative evaluation of the EMG, and other artifacts. Based on previous studies, it is well known that the hemodynamic response to a brief period of neural activation typically peaks 4 or 5 s after the neural response. Taking this into account, we planned to analyze the period ranging from $-2.0$ to $+6.0\mathrm{s}$ after EMG onset for the NIRS data analysis. Therefore, when the interval was under 9 s, the epochs were removed. The same procedure was applied if the EMG recording was unsuitable (e.g., very low amplitude or lacking a single movement). Finally, artifact-free epochs were averaged according to EMG onset with a baseline calculated from 2000 ms to 1600 ms before EMG onset.

2.5.

NIRS Data Acquisition

The standard method for NIRS has been described in detail elsewhere.^2,6 NIRS was carried out with a $695/830\mathrm{nm}$ spectrometer (ETG-4100; Hitachi Medical Corporation, Japan), using $4\times 4$ probe sets (24 channels for each hemisphere). The inter-probe distance was 3 cm, and NIRS signals were recorded at a sample rate of 10 Hz from all channels. The trigger mark was obtained from the EMG signal as described in the MRCP data acquisition section (Sec. 2.3).

2.6.

NIRS Probe Placement Procedure

As shown in Fig. 2(b), the position of each channel was numbered from 1 to 48. In order to cover the bilateral motor cortex and motor association cortex, the probes were placed symmetrically beside Cz since C3 and C4 were enclosed by channel numbers 16, 19, 20, and 23 and numbers 40, 43, 44, and 47, respectively. Consequently, Fz was placed essentially between channel numbers 4 and 28 in all subjects. In order to confirm the precise positions of the probes, we demonstrated a probe-cortex correlation using an appropriate software system (EZT-DM 101/102; Hitachi Medical Corporation, Japan). In this system, the spatial information of each probe on the head surface was first registered. Second, the 3-D head surface image and the 3-D brain image were generated from the subject’s own T1-weighted MRI data (MAGNETOM Verio 3T, Siemens, Germany), using 3-D reconstruction software (ZedView, LEXI, Tokyo, Japan), and each probe position was mapped onto the 3-D head surface image. Finally, probe positions were superimposed onto the subject’s own 3-D brain image [Fig. 2(c)]. In the present study, this procedure was performed in all subjects, and we obtained accurate regional information about hemodynamic responses to neural activation. Figure 2(c) shows that the bilateral prefrontal cortex (PFC), the bilateral dorsal premotor cortex (dPMC), and the bilateral hand area of the SMC were covered by the placement of the probes. According to the results of these processes, the regions of interest (ROIs) were set in all subjects for PFC (channels 4, 5, 28, and 29), dPMC (channels 11, 12, 35, and 36), and the hand area of SMC (channels 16, 19, 20, 23, 40, 43, 44, and 47) within an acceptable structural error range due to variable head or brain shape, as shown in Fig. 2(b).

2.7.

NIRS Data Analysis

The data analysis was performed using a custom-made software program (MATLAB; The Math Works). The epochs that had been removed from the MRCP data analysis were excluded, and the remaining data were filtered by a band-pass filter with a range of 0.08 to 0.75 Hz. The average baseline Hb concentration was calculated for the interval from 2.5 s to 2.0 s before EMG onset, and the concentration changes were analyzed from the pre-movement phase (2.0 s before EMG onset) to the post-movement phase (6.0 s after EMG onset). That is, each period ranged from $-2.0\mathrm{s}$ to $+6.0\mathrm{s}$ after EMG onset. During this interval, the relative concentration changes of ${\mathrm{HbO}}_2$, Hb, and HbT were averaged, and the time courses were obtained for each channel.

2.8.

Comparison Between MRCP and NIRS Data

In order to evaluate the coupling between electrophysiological neural activity and hemodynamic changes, the averaged waveforms of the MRCPs at Fz, Cz, C3, and C4 and the averaged ${\mathrm{HbO}}_2$, Hb, and HbT changes in the bilateral PFC/dPMC and the hand area of the contralateral SMC were obtained. These results allowed time data comparisons between MRCP and NIRS.

3.1.

Behavioral Data

The mean interval between pinch movements was $10.2\pm 0.8\mathrm{s}$, and the average number of movement trials subjected to analysis was $89\pm 4$. We excluded 13% of trials (shorter interval, 9%; very low amplitude EMG, 1%; lacking a single movement, 1%; artifacts, 2%). All subjects were cooperative during the entire session.

3.2.

MRCP Results

Figure 3 shows the averaged MRCP waveforms for subject one during the performance of a voluntary right thumb-index finger pinching movement conducted at his own pace. In subject one, the initial slow negativity, known as the Bereitschaftspotential (BP),⁹ started 1590 ms before EMG onset and peaked 905 ms before EMG onset. Following the BP, the steeper component with a negative slope (${\mathrm{NS}}^{\prime }$),¹⁵ occurred 535 ms before EMG onset and terminated 45 ms after EMG onset. Maximal BP was observed at Cz, while maximal ${\mathrm{NS}}^{\prime }$ was detected at C3. The motor potential (MP),¹⁶ following ${\mathrm{NS}}^{\prime }$ was then recognized mainly at C3, and it peaked 125 ms after EMG onset. The averaged MRCP values, including BP, ${\mathrm{NS}}^{\prime }$, and MP, were clearly recognizable in subjects two and three, as well as in subject one. The latency and the maximal position of the MRCP components for each subject are presented in Table 1. The maximal BP peak was observed in Cz or Fz, while ${\mathrm{NS}}^{\prime }$ or MP showed significant maximums mainly at C3 for all subjects.

Fig. 3

Average MRCP waveforms at Fz, Cz, C3, and C4, and the EMG recording for subject one. The vertical dotted line marks the onset of the EMG burst. The Bereitschaftspotential (BP) was maximal at Cz, while the negative slope (${\mathrm{NS}}^{\prime }$) was maximal at C3. The following motor potential (MP) was dominant at C3. Similar results were obtained for all subjects.

Table 1

Latency and maximal position of the MRCP components in each subject.

3.3.

NIRS Results

The average changes of ${\mathrm{HbO}}_2$, Hb, and HbT compared to baseline for subject one are shown in Fig. 4. In subject 1, the PFC and dPMC were bilaterally and symmetrically activated with similar amplitudes before EMG onset. Following these activations, the hand area of the SMC showed left-dominant activation and a peak after EMG onset. Figure 5 shows the average changes of ${\mathrm{HbO}}_2$, Hb, and HbT in each ROI for each subject, and the time to peak for the average ${\mathrm{HbO}}_2$ changes in each ROI for subjects one, two, and three typical responses. The time to peak for the average ${\mathrm{HbO}}_2$ changes in each ROI for each subject is presented in Table 2. For all subjects, the PFC and dPMC were bilaterally activated before EMG onset. The cross-correlation coefficient from each subject between the two hemispheres’ NIRS signals ranged from 0.92 to 0.98 in the PFC, and ranged from 0.94 to 0.98 in the dPMC. Thus we averaged the bilateral temporal PFC and dPMC data and considered the averaged data to be identical for the left and right. The ${\mathrm{HbO}}_2$ peak in the PFC preceded that in the dPMC by 0.1 s or occurred at the same time. The mean peak latency of ${\mathrm{HbO}}_2$ was $0.6\pm 0.1\mathrm{s}$ before EMG onset in the PFC and $0.5\pm 0.1\mathrm{s}$ before EMG onset in the dPMC. Following these activations, the hand area of the SMC showed left-dominant activation $3.7\pm 0.2\mathrm{s}$ after EMG onset. The topography of the average ${\mathrm{HbO}}_2$ changes in subject three showed that hemodynamic activation occurred earlier in the bilateral PFC and dPMC than in the hand area of the SMC, with left dominance (Fig. 6).

Fig. 4

Average changes of ${\mathrm{HbO}}_2$, Hb, and HbT compared to baseline in response to a simple right-handed pinching task in subject one. Before EMG onset, the PFC and dPMC were bilaterally and symmetrically activated with similar amplitudes. The hand area of the SMC significantly activated after EMG onset with contralateral dominance.

Fig. 5

Typical hemodynamic responses in each ROI for each of three individual subjects. For all subjects, PFC and dPMC were bilaterally activated before EMG onset, and the peak time of ${\mathrm{HbO}}_2$ was the same ($<0.1\mathrm{s}$) for these two regions with similar amplitudes. Following these activations, the hand area of the SMC showed left-dominant activation after EMG onset. The error bars indicate standard errors for ${\mathrm{HbO}}_2$ and Hb.

Fig. 6

The topography of the averaged ${\mathrm{HbO}}_2$ changes in subject three. The hemodynamic activation that occurred earlier in the bilateral PFC and dPMC than in the hand area of the SMC showed left-dominance.

Table 2

Time to peak of averaged HbO2 Changes for each subject in PFC, dPMC, and the contralateral hand area of SMC.

3.4.

Results of Comparisons Between MRCP and NIRS Data

Figure 7 presents the average time-series waveforms of MRCP and NIRS signals in the PFC, dPMC, and contralateral hand area of the SMC in subject three. Similar results were obtained in the others. For all subjects, the onset of MRCP was followed by the hemodynamic responses. The sources of MRCP were various cortical projection neurons in the motor association areas.¹⁷ Thus MRCP may have been contaminated by various active potentials from these regions.¹⁵ According to previous MRCP study with MEG,¹⁸ the motor field that peaks at EMG onset is present in the bilateral primary motor areas with contralateral dominance, and the movement-evoked field I, which is detected about 0.1 s after EMG onset, is present only in the contralateral somatosensory area. Since both are thought to be counterparts of the MP component of MRCP,¹⁹ the MP component in C3 can be regarded as regional potentials mainly in the contralateral SMC. Taking these factors into account, we regarded the peak ${\mathrm{HbO}}_2$ increase after EMG onset in the hand area of the SMC as the response to neural activation in the same region. Hence, in order to obtain temporal information about the vascular response to neural activity, we measured the peak time of the MP component in C3 (Table 1) and the peak time of ${\mathrm{HbO}}_2$ changes in the contralateral hand area of the SMC (Table 2) and then calculated their difference for each subject. The difference was $3.6\pm 0.2$ on average.

Fig. 7

Average time-series waveforms of MRCP and ${\mathrm{HbO}}_2$ changes in the mPFC, mPMC, and the contralateral hand area of the SMC in subject three. The onset of MRCP was followed by hemodynamic responses. The ${\mathrm{HbO}}_2$ increase after EMG onset in the contralateral hand area of the SMC was regarded as the vascular response to the MP component in C3.