Working Memory Event Encoding By Theta And Gamma Phase Oscillation Of Multi-channel LFPs Based On Wigner-Ville Distribution

ObjectiveThis work studied phase synchronous oscillation modes and encoding for working memory event in 0 andγbands of multi-channel local field potential (LFPs) based on Wigner-Ville Distribution (WVD). This work studied the various components of a bandwidth of 2Hz phase synchronous mode of working memory on the y band. Experimental data were 16-channel LFPs acquired from prefrontal cortex of rats during working memory task related to the event 0-7s. The collaborative coding pattern provides neural computation support for analyzing the neural coding mechanisms of working memory.Methods1. Experimental data:the 16-channel raw experimental data in this paper are provided by our laboratory. Rats are trained in the Y maze of working memory task firstly. When the rats correct behavior reached more than 80%, the rats are considered to master working memory task. And then 16-channel microelectrode array are implanted in the rat prefrontal cortex. The United States Cyberkinetics's Cerebus-128 data acquisition system is used to record the 16-channel raw data of the four SD rats on prefrontal cortex in Y maze working memory task. The study selected 7s data sufficient to the whole process of working memory.2. Preprocessing of original data:the 16-channel raw data are filtered with low-pass filter to obtain the 16-channel LFPs (0.3-500Hz). The each channel LFP was fir to obtain the fitting curve and then original LFP minus the fitting tendency. The zero-baseline 16-channel LFPs are obtained by baseline fitting after removing the interference of power frequency.3. Time-frequency analysis was used to LFP (FFT). The energy concentrated on 40-60Hz during the working memory.4. Obtaining different rhythm components:Bandpass filter is used to decompose the 16-channel LFPs to different bands.5. The instantaneous phase extraction of 16-channel LFPs:the Hilbert transform was used to extract the instantaneous phase. 6. Reference channel:select the highest average firing rate of the channel as a reference channel.7. Sliding window and step parameter:The length of sliding widow is 50ms and step is 12.5ms.8. Multi-channel signals dynamic WVD analysis:(1)WVD analysisStarting from the initial window, the WVD was used to calculate in each window between channel phase sequence of LFPs'different bands and reference channel phase sequence over a sliding 50 ms multi-taper window, with 12.5ms overlapping. Then the nonlinear terms of LFPs'different bands were obtained. The WVD was also used to calculate phase sequence of LFPs'different bands in each window to obtain linear terms of LFPs. The nonlinear and linear terms of spatial and temporal distribution patterns of synchronous oscillation are obtained. Then we study the LFPs synchronous oscillation and encoding of working memory.(2) The wavelet packet transformThe Hilbert transform is fit for narrowband signal. In order to verify the validity of the results of broad physiological bands, this paper apply WVD analysis to the key bands which encode the working memory events further.Wavelet packet analysis is used to decompose the y band (30-80 Hz) for 9-layer. then we got the 2Hz bandwidth of LFPs. y band corresponding to the 16-40 packets. The corresponding bands are 30-32Hz-78-80Hz. The nonlinear WVD is used to analyze the wavelet packet component. Starting from the initial window, we calculate in each window between channel phase sequence of LFPs'and reference channel phase sequence. Then the nonlinear and linear terms of LFPs'every different 2Hz range on y band were obtained. The nonlinear and linear terms of spatial and temporal distribution patterns of synchronous oscillation are obtained.ResultsThis paper uses the nonlinear WVD to analyze the 16-channel LFPs andθ,γphase synchronous mode in the Y maze test of working memory, and studys the refinement y band phase synchronization mode with the bandwidth 2Hz component. The main results are as follows: 1. During the working memory event, multi-channel LFPs'energy concentrates on the 40-60Hz.2. During the working memory event, multi-channel LFPs'nonlinear WVD values in various physiological bands before reference point 2s in 4 rats with10 experiments.Through t test, multi-channel LFPs'nonlinear WVD terms exsit obvious synchronous oscillation on the 0 and y bands (P<0.05). The nonlinear WVD items on d, a, and B-band do not exsit synchronous oscillation (P>0.05).3. During the working memory event, multi-channel LFPs'linear WVD values in various physiological bands before reference point 2s in 4 rats with 10 experiments. Through t test, multi-channel LFPs'linear WVD terms exsit obvious synchronous oscillation on the 0-band (P<0.05). The linear WVD items on the other bands do not exsit synchronous oscillation (P>0.05).4. During the working memory event, multi-channel LFPs'nonlinear WVD values on 40-42Hz-58-60Hz before reference point 2s in 4 rats with 10 experiments.Through t test, multi-channel LFPs'nonlinear WVD terms exsit obvious synchronous oscillation on 40-42FL,44-46Hz,46-48Hz,54-56Hz,56-58Hz,58-60Hz bands (P＜0.05). The nonlinear WVD items do not exsit synchronous oscillation on 42-44Hz,48-50Hz,50-52Hz,52-54Hz (P>0.05).5. During the working memory event, multi-channel LFPs'linear WVD values on 40-42Hz-58-60Hz before reference point 2s in 4 rats with 10 experiments. Through t test, multi-channel LFPs'linear WVD terms exsit obvious synchronous oscillation on 44-46Hz (P<0.05). The linear WVD items on the other bands do not exsit synchronous oscillation (P>0.05).Conclusions1. During the working memory event, multi-channel LFPs'energy concentrates on the 40-60Hz which is closely related with the working memory event.2. During the working memory event, nonlinear WVD is applied to analyze LFPs and find obvious synchronous oscillation on 0 and y bands which encoding the working memory event efficiently. The nonlinear WVD items do not exsit synchronous oscillation on d,a,and B bands which do not encode the working memory event. Linear WVD is applied to analyze LFPs and the obvious synchronous oscillation exist onθ-band and do not exsit synchronous oscillation onγ-band. It shows that the WVD analysis is effective.3. During the working memory, nonlinear WVD is applied to analyze LFPs and the obvious synchronous oscillation exsit on 40-42Hz,44-46Hz,46-48Hz,54-56Hz, 56-58Hz,58-60Hz, which encoding the working memory event efficiently. The nonlinear WVD items do not exsit synchronous oscillation on 42-44Hz,48-50Hz, 50-52Hz,52-54Hz which do not encode the working memory event. Linear WVD is used to analyze LFPs.The obvious synchronous oscillation exists on 44-46Hz which encoding the working memory event efficiently and do not exsit synchronous oscillation on other bands. It shows that the WVD analysis is effective.