A. Optical recording
The optical signals acquired by the CW-NIRS were converted into concentration changes of oxygenated and deoxygenated hemoglobin (Δ[HbO2] and Δ[HHb] respectively) according to the theory described in our previous paper . Data sets for each volunteer were analysed individually and expressed in micromolar concentrations (μM). The signals to be analysed were detrended to remove baseline drifts. The raw data clearly showed a marked increment of the response amplitude when the stimulus passed from the grey to windmill pattern; another relevant aspect was the feature related to heartbeat and other low frequency fluctuations (less than 1 Hz). However, no clear fast response could be observed in any subject.
To detect possible higher frequency components (starting from 8 Hz, the reversal frequency of stimulation) a non-parametric FFT algorithm was used on the [HbO2] signals. The analysis was extended to only Δ[HbO2] since previous studies show that spontaneous fast oscillations are most prominent in the [HbO2] signal . We transformed two time windows of 40 seconds: the first located in the first rest period, whereas the second was chosen during stimulation at maximum contrast.
Figure 5 shows the oscillatory changes in [HbO2] together with the corresponding power spectral analysis for one subject, typical of the whole group.
The power spectrum signal shows relevant components in a variety of frequency bands. There is a pronounced peak at the heart rate frequency around 1 Hz (P-waves), a broad peak at the breathing rate around the range at 0.3–0.4 Hz (R-waves) and a peak in the range around 0.1 Hz (M-waves) .
Furthermore, harmonic components that have coherent properties with these signals can be observed. This distribution of slow hemodynamic trends indicates that light penetrates into cortical tissue .
Around 8 Hz both spectra were essentially featureless, thus, as reported by other researchers , we did not observed the fast optical response to the windmill stimulation. However, as discussed by Franceschini and Boas , physiological fluctuations such as arterial pulsation caused intensity changes 300–1500 times larger than those expected from the fast signal. In Figure 5, the ratio between the noise floor at 8 Hz and the arterial pulsation changes is about 1/100, thus it is reasonable that no evidence of the fast signal is present in our spectrum. Nevertheless, exploiting the high sensitivity of our instrument, this controversial aspect could be investigated by averaging several dozen trials obtained by synchronization between stimulus and epoch acquisition.
On the other hand, slow hemodynamic changes are well visible in all subjects tested.
As an example, Figure 6 shows the signals acquired from one of the subjects as representative of typical Δ[HbO2] and Δ[HHb] responses obtained from the occipital cortex. The raw data was filtered using a moving average filter of 1.8 seconds to reduce the oscillatory changes reported in Figure 5, i.e. mainly the heart and breathing rate. As expected, the behaviour of oxygenated haemoglobin concentration reflects an increased activity of the visual cortex during the visual stimulation period. In the figure, each curve corresponds to the variation of concentration as acquired by the NIMO system using different stimulus contrasts, i.e. 1%, 10% and 100%. As shown previously in fMRI experiments , the hemodynamic response increases with stimulus contrast; this boost is directly dependent on the analogous bust in neural activity.
Let us consider, for example, the results obtained at maximum stimulus contrast: the recorded signal presents a reasonably stable baseline corresponding to the first 60 second non-stimulation period and variations of oxyhemoglobin concentration during stimulation. The b/w windmill pattern stimulus induced an increase in Δ[HbO2] concentration of about 3 μM. A rapid rise of the signal was observed after a short delay and then a stable level was maintained for the duration of the stimulus. [HbO2] concentration returned to baseline after about 10 seconds. Figure 6 shows that the Δ[HHb] time-course is quasi-complementary and has a very similar time scale. After an initial stable baseline, [HHb] decreased in response to stimulus in about 10 seconds. The lower concentration level remained constant up to the end of the stimulus, when the baseline condition was restored within 10 seconds.
The measuring results recorded on the pooled subjects showed a similar trend: the concentration of [HbO2] increased and [HHb] decreased after the onset of stimulation. As expected from previous studies , [HbO2] increased at least twofold the magnitude of the decrease in [HHb]. Hemodynamic response occurs within a few seconds after visual stimulus begins. For all subjects considered, the response reaches its maximum about 20 seconds after stimulus onset. The delay and the rise time of the signals were not apparently related to stimulus contrast. In some recordings the signals reach a plateau and decline after the stimulation period. The time course of the changes recorded is consistent with previously reported vascular responses over the occipital region [27,28].
B. Electrical recording
To quantify neural activity during resting and stimulation conditions, we monitored the VEP signals. The alternation of black and white checkered pattern of the stimulus induced a corresponding fluctuation of the VEP signals, acquired by averaging the EEG recordings. A null response, as usual, was recorded when the subject was looking at the screen displaying a uniform grey image with the same total luminance as the patterned stimulus; this response was used as comparison with those obtained after stimulation.
Typical VEP signal responses are shown in Figure 7. Two different trends can be observed: during the rest condition (grey screen) there is no prominent frequency component and the VEP amplitude is roughly null. During the stimulation period the VEP signal presents a major oscillation at twice (16 Hz) the reversal trigger frequency (8 Hz): as a general rule, the retina is stimulated by pattern variations and our stimulus gives two "black and white inversions" for every triggering period. The VEP response, accordingly, presents two main oscillations, thus confirming the correlation between electrical activity and visual stimulation. The VEP amplitude increases with contrast and, despite the figure is referred to a single subject, it reflects the general property of the vision system. Similar results have been obtained in all subjects monitored. The correspondence between visual stimulation protocol and recording duration, which lasts for 60 seconds is evident in Figure 7. Each 60-second interval corresponds to the interval reported as a white/grey vertical strip in Figure 6.
As shown previously
], VEP responses increased with stimulus contrast; this increase was directly attributable to the correspondingly increased neural activity.