hide
Free keywords:
-
Abstract:
Introduction:
The interpretation of the neuroimaging signal is rendered difficult by the fact that it has been reported to correlate with different neuronal processes, such as multi-unit spiking, and activity in the various frequency bands of the LFP [1]. Spikes and low-frequency LFPs may, however, represent different neural processes, such as features of a visual stimulus or neuro-modulatory inputs [2]. This makes the inability to differentiate between contributions of spiking and low-frequency LFP activity a significant problem in the interpretation of hemodynamic signals. We developed a novel technique to study local neurovascular coupling enabling simultaneous epidural functional near-infrared spectroscopy (fNIRS) and intra-cortical electrophysiology, and aimed to determine if different neuronal processes had different correlates with hemodynamic signals.
Methods:
We recorded fNIRS together with multiple microelectrode recordings from a small volume of primary visual cortex in two anesthetized monkeys. Both stimulus-induced and spontaneous activity were recorded. Each visual stimulation trial consisted of 5s of a whole-field rotating chequerboard (ON) followed by 15s of a blank screen (OFF). Spontaneous activity was recorded for 15 min per run, in the absence of visual stimulation, and subject's eyes closed.
The electrophysiological signal was filtered into eight frequency bands, (namely DeltaTheta (1-8 Hz), Alpha (9-15), Spindle (15-20), low Gamma (20-40), Gamma (40-60), high Gamma (60-120), very high Gamma (120-250) and MUA (1-3k)), and their band envelopes obtained. Multi-unit spike-rates were obtained by counting number of spikes in 50ms bins. Visual modulation for each band was obtained by the forumula: ON-OFF/ON+OFF; where ON and OFF are the band powers during ON and OFF epochs, resp. Results:
We observed a negative correlation between the spiking activity and hemodynamic response peak-amplitude, but a positive correlation with its peak-time. Specifically, the total spike count during the ON epoch correlated strongly with the HbO peak-time. The peak-spike rate during the ON epoch also correlated strongly with the HbO initial-dip, demonstrating that it reflects bursts in spiking activity. We observed that the HbO peak-amplitude correlated strongly with modulations in the DeltaTheta and Alpha bands. All these relationships were also observed in spontaneous activity, demonstrating that they do not arise due to strong visual stimulation.
We also found a strong difference in the spatial spreads of low versus high-frequency activity. Although we used whole-field visual stimulation, modulations in spike-rates were more spatially localized, but less synchronous than those in the DeltaTheta band. With more synchronous arteriole recruitment, stronger low-frequency LFPs lead to larger HbO response amplitudes. In contrast, vascular responses to spiking integrate temporally. With longer durations of dilation, peak-time is affected more than peak-amplitude.
We also observed strong negative correlations between the DeltaTheta modulations and spiking activity. While the DeltaTheta band wasn't visually modulated, it was Conclusions:
Briefly, low-frequency LFPs are reflected in hemodynamic peak-amplitude, and high-frequency activity in its peak-time. These results help better understand the mechanisms underlying neurovascular coupling, and enable better interpretations of hemodynamic signals observed during functional neuroimaging studies. anti-correlated with spiking, supporting the notion that high and low-frequency LFPs represent different neuronal processes. Further analysis based on system-identification also showed that the relationship between fNIRS signals and underlying neuronal activity are similar to those reported for fMRI [3].
