Fig. 9.

Comparison of NIRS data with model predictions for the subject S2, whose data did not meet the criteria the functional activation. Panel layout is the same as for Fig. 8. The ΔHbO₂ signal is well-explained by both scalp and brain models. The ΔHHb response is marginally positive, so may be better explained by the scalp model, although the scale of this response in all cases is small relative to the signal variability and the correlations are poor. The modelling is not conclusive, but it does suggest that the negative classification in this case may be false.