Decoding the neural activity of dorsal spinal nerve root evoked by acupuncture at Zusanli point based on the generalized linear model

Abstract

Neural system can characterize the information evoked in external stimulations by spatiotemporal encoding. Acupunctureas, as a mechanical actionis equivalent to an external stimulus to the neural system. In order to decipher how neural systems represent and transmit acupuncture information, experiments are designed such that different frequencies of manual acupuncture (MA) manipulations can be given to ‘Zusanli’ point of an experimental rat, and the induced electrical signals in spinal dorsal root ganglion are detected and recorded. First, the spike trains are obtained by data preprocessing and transformed into point process trains. Then a generalized linear model is introduced to encode the acupuncture information and produce the simulated neural spike train. In addition, a Bayesian decoding algorithm is used to reconstruct the displacement oscillograms of the acupuncture needle from the neural point process spike train. Finally, the quantile-quantile (Q-Q) plot based on the time-rescaling theorem is used to evaluate the agreement between the encoding model and the point process data. Results prove to be valid based on statistical analysis. These studies have offered new insights into neural processing underlying acupuncture and have implementations for constructing the interface between neural systems and machines and improving the clinical study.