Fast optical signal in visual cortex: Improving detection by General Linear Convolution Model

In this study we applied the General Linear Convolution Model to fast optical signals (FOS). We modeled the Impulse Response Function (IRF) as a rectangular function lasting 30 ms, with variable time delay with respect to the stimulus onset. Simulated data confirmed the feasibility of this approach and its capability of detecting simulated activations in case of very unfavorable Signal to Noise Ratio (SNR), providing better results than the grand average method. The model was tested in a cohort of 10 healthy volunteers who underwent to hemi-field visual stimulation. Experimental data quantified the IRF time delay at 80-100 ms after the stimulus onset, in agreement with classical visual evoked potential literature and previous optical imaging studies based on grand average approach and a larger number of trails. FOS confirmed the expected contralateral activation in the occipital region. Correlational analysis between hemodynamic intensity signal, phase and intensity FOS supports diffusive rather than optical absorption changes associated with neuronal activity in the activated cortical volume. Our study provides a feasible method for detecting fast cortical activations by means of FOS.

► General Linear Convolution Model was applied to fast optical signals. ► The method works better than grand average based models. ► Visual cortex fast responses were detected even at single-subject level. ► Results agree with visual evoked potential and fast optical literature. ► Diffusive properties of optical cortex change with neural activity.