The overall purpose of this paper is to combine optical and electrical recordings. Based on preliminary analyses of individual averaged waveforms, we consider the average variations in [HbO2], [HHb] and VEPrms that occur during stimuli and their correlation.
Filtered NIRS signals are considered to evaluate the variations of [HbO2] and [HHb] due to visual stimuli. The signal variation observed during each stimulation is calculated as the difference of the average signal during the stimulus and the average baseline recorded during the rest period. The variations observed during the four epochs were then averaged to obtain 〈Δ[HbO2]〉 and 〈Δ[HHb]〉 for each subject.
The EEG variations due to stimulus response is very small, has a poor signal-to noise ratio and cannot be directly extracted and measured from the EEG signal. This happens at every contrast value, so an averaging technique is required to record a reliable VEP. A 60-second stimulation at 8 Hz produces about 400 epochs to be averaged if an "amplitude window" is applied to the EEG signal in order to discard epochs containing excessively large amplitude variations.
All the averaged signals showed reliable trends and good signal-to-noise ratio, and they were accepted for the statistical analysis, allowing the VEPrms calculation for each stimulus.
The mean pooled hemodynamic variations for each stimulus contrast, averaged over observers, are shown in Figure 8. The standard errors are bounded by 0.15 μM and 0.34 μM for 〈Δ[HbO2]〉 and 0.04 μM and 0.06 μM for 〈Δ[HHb]〉. As shown in the figure, a logarithmic trend of hemodynamic response as a function of stimulus contrast is observed. Strong linear correlation (R = 0.998 and 0.997 for 〈Δ[HbO2]〉 and 〈Δ[HHb]〉, respectively) is found between 〈Δ[HbO2]〉 and 〈Δ[HHb]〉 changes and the logarithm of contrast.
Figure 9 shows the mean electrical signal detected 〈 VEPrms〉 for each stimulus contrast, averaged over observers. The standard errors are bounded by 0.18 μV and 0.20 μV. Also in this case we observed a strong linear correlation (R = 0.994) between 〈 VEPrms〉 and the logarithm of contrast.
There was a linear relationship between hemodynamic response and VEP response during graded visual stimulation: indeed, strong linear correlation was found between 〈Δ[HHb]〉 and 〈 VEPrms〉, as shown in Figure 10 (R = 0.988 and 0.984 for and 〈Δ[HHb]〉, respectively).
Even if the number of subjects is not sufficient to run statistical analysis, the small intrasubject standard deviation suggests that all subjects might have, in the visual cortex, roughly a similar electrical and hemodynamic response to the windmill pattern stimulation.
Neither NIRS nor electrical response is a linear function of contrast, but both increase monotonically with stimulus contrast. A "logarithmic" type of nonlinearity is evident for both the responses. Our study investigated simultaneous effects of visual contrast on hemodynamic and VEP responses. We found a strong correlation between visual log contrast and NIRS signals as well as between log visual contrast and VEP. To our knowledge, the present study is the first to address the issue of simultaneous assessment of NIRS and VEP during graded contrast visual stimulus. Nevertheless, our results are consistent with other studies, which have explored the effect of visual contrast on VEP [21,29,30]. The 'logarithmic' hemodynamic response that we observed is also in agreement with previous studies using Doppler  and fMRI BOLD .
It seems that both hemodynamics and VEP show saturation behaviour at high visual contrasts. This result is supported by animal studies of primary visual cortex neurons, which have shown that the contrast response function of spike rate responses has a sigmoidal shape. In general, neurons show an increasing response that is nonlinear at low contrasts, and a linear increase up to a certain contrast level at which the response reaches its asymptote [32-34].
Like the neural responses, fMRI BOLD signal was also found to be a nonlinear function of stimulus contrast ; however, a linear system analysis on the fMRI responses predicted a linear relationship between the hemodynamics and neural activity , as observed in our study.
LR and GS carried out all the optical measurement setup and drafted the manuscript; LB and SF provided the stimulation and electrical recording software. All the authors contributed to refining the paper's content and structure. SF took care of revision according to peer review and all authors read and approved the final manuscript.
The authors wish to thank Ing. M. Massari for his contribution during the experimental activities, Prof. R. Ferrari, Prof. G. P. Biral and Ing. M. Corradini for their valuable suggestions, and gratefully acknowledge the volunteers who gave consent to be involved and monitored during visual stimulations. Research partially supported by grant n.MM09163913_004 from MIUR, Italy.