Machine-learning based automatic and real-time detection of mouse scratching behaviors

摘要

Analysis of neural signals recorded from substantia nigra pars compacta for brain machine interface in freely moving rat There have been various studies using bio-signals in machine control research for paralyzed patients. Among them, it has been shown that some of the machines could be controlled by using the neural signals obtained from the muscles and the motor cor- tex in relation to the movement. However, previous studies have been required a number of electrodes to acquire and analyze var- ious signals from each segmented motion domain. To overcome these drawbacks, in this study, we used fewer electrodes to record and analyze the signals related to the movement of rat in the sub- stantia nigra pars compacta (SNc). Electrodes (single-channel) were implanted in bilateral SNc regions of male Sprague-Dawley rat, and modeling of rat movement was determined through electrical stim- ulation. In order to confirm the net neural response related to the motion of rat in the non-anesthetic state, the noise signal generated in the general motion was compared with the signal generated in the rotational motion. Through this procedure, it was filtered the noise signal generated in the general motion. The results for this study were such as following: First, the specific firing rate was not observed from analyzing the neural signal in the anesthetic state. Second, analysis of neural signal in the non-anesthetic state was shown that the firing rate in the left SNc region was higher when the rat move to the right. On the other hand, the gamma band changes of the local field potential in the right SNc region was larger when moving to the left. These results show that the neural signals of the SNc region related to the rat movement can be used as input signals to control the machine, especially in the brain machine interface. Furthermore, it is expected to be helpful for patients with paral- ysis and nerve damage. This study was supported by the grant from CABMC (Control of Animal Brain using MEMS Chip) funded by Defense Acquisition Program Administration (UD140069ID).