| Orientation selectivity plays a very important role in the process of vision formation.Neural oscillations exist widely in different brain regions.Coupling occurs between neural oscillations in different frequency bands,resulting in cross-frequency coupling.Studies have shown that cross-frequency coupling is an important foundation for brain function.In this paper,implanted microelectrode arrays were used to collect local field potential signals in V1 and V4 regions of macaque visual cortex,the cross-frequency coupling between low-frequency and high-frequency neural oscillations under different orientation stimuli was analyzed.Then we use Lasso logistic regression model to build the prediction model of grating orientation.This work provides a new idea for exploring the formation mechanism of neuronal orientation selectivity.First,the data collected by the microelectrode array is preprocessed to obtain the phase and amplitude of low-frequency and high-frequency neural oscillations.The electric signal was low-pass filtered and band-pass filtered using a fourth order Butterworth band-pass filter to obtain low-frequency signals in the 0-30 Hz and 4-12 Hz bands,and high-frequency signals in the 30-200 Hz and 30-90 Hz bands.Then the Hilbert transform is used to extract the phase and amplitude of the filtered signals.Secondly,the phase-amplitude coupling based on phase-locking value and generalized eigendecomposition,amplitude-amplitude coupling based on Pearson correlation coefficient were used to analyze the cross-frequency coupling strength under different orientation stimuli.We found that its changes regularly with the change of orientation stimuli.Circular statistics were used to quantify the grating tuning curve.It can be seen from the results that phase-amplitude and amplitude-amplitude cross-frequency coupling can characterize orientation selectivity.In addition,the preferred orientations of neurons in the V1 and V4 brain regions are analyzed,and it can be seen that the optimal orientation distribution obtained by cross-frequency coupling based on generalized feature decomposition is relatively concentrated,which can effectively determine the optimal orientation of neurons.Finally,based on the Lasso logistic regression model,the strengths of phase-amplitude coupling,the strengths of amplitude-amplitude coupling,and the joint features of the two are used to establish a regression model between the neuron cross-frequency coupling strength and the grating orientation.We implemented the prediction of grating orientation by cross-frequency coupling strength,and used the cross-validation method to analyze the accuracy of the prediction.It was found that different types of cross-frequency coupling and preference orientation of neurons affect the accuracy of prediction model. |
