猕猴运动皮层中局部场电位信号的长期稳定性及其同侧运动解析
发布时间：2018-08-24 08:04
【摘要】：物理损伤或神经系统疾病等原因经常会导致人们永久性地丧失运动功能,而脑机接口技术的出现给这些残障患者带来了新希望。脑机接口直接提取大脑的运动信息,绕过常规的脊髓和外周神经系统,直接控制外部设备,为运动功能重建提供了一种新思路。目前植入式脑机接口中常用的神经信号主要有两种：神经元锋电位信号和局部场电位信号。其中神经元锋电位信号,具有较高的时空分辨率,可以携带较为丰富的运动信息,但是在实际应用中难以维持较长时间的有效的记录。近年来诸多研究表明局部场电位信号中也蕴含着丰富的运动信息,但其长期稳定性相对研究较少。另外,不同频段的局部场电位信号对同侧肢体运动的信号表征及其解析情况也尚不十分清楚。 本研究基于猕猴的脑机接口平台,在三只猕猴进行四方向center-out实验时,使用Blackrock的硅阵列电极同步记录猕猴初级运动皮层和背侧运动前区中的神经信号,从神经信号特征、神经信号与运动信号间的相关性以及神经信号解码三个方面对局部场电位信号的长期稳定性及其对于同侧运动信息的解析进行了分析和研究。结果表明在长期稳定性方面,相对于神经元锋电位信号,局部场电位信号记录质量衰减缓慢,能够维持相对更久的运动方向信息。在神经元锋电位信号记录质量较差时,局部场电位信号能够表现出更好的解码性能。特别地,在整个神经电极阵列记录不到神经元锋电位信号时,局部场电位信号仍然能够提供一定的解码信息。在同侧运动表征与解析方面,多个频段的局部场电位信号表征了同侧肢体的运动信息。同时,同一神经信号表现出与对侧运动时显著不同的偏好方向。解码结果表明多个频段能够精确解析同侧肢体的运动信息,其中以200-400Hz频段的解码性能最佳。 本研究工作分析了神经元锋电位信号记录质量较差时的局部场电位信号的解码性能,并对不同频段的局部场电位信号对同侧肢体运动的信息表征及其解析进行了深入研究。主要创新点在于：(1)在四方向center-out运动范式上,从信号特征、方向性调谐和离线解码性能三个方面证实了局部场电位信号具有比神经元锋电位信号更好的长期稳定性；(2)研究了神经元锋电位信号记录质量在严重衰减甚至完全消失时,局部场电位信号在四方向分类解码上仍具有一定的生理意义；(3)研究表明多个频段的局部场电位信号蕴含同侧肢体运动的有效信息,尤其在200-400Hz频段信号的解码性能相对较好,四方向的分类正确率最高。 本文通过对局部场电位信号的长期稳定性及其对于同侧肢体运动信息表征和解析的研究表明,大脑运动皮层的局部场电位信号同样蕴含丰富的肢体运动信息,可以作为一种稳定的神经信号源,相关研究结果将进一步促进和增强植入式脑机接口的临床应用。
[Abstract]:Physical injury or nervous system diseases often lead to permanent loss of motor function, and the emergence of brain-computer interface technology brings new hope to these disabled patients. Brain-computer interface directly extracts motor information from the brain, bypasses the conventional spinal cord and peripheral nervous system, and directly controls the peripheral equipment, which provides a new idea for the reconstruction of motor function. At present, there are two kinds of nerve signals commonly used in implantable brain-computer interface: neuron spike signal and local field potential signal. The neuron spike signal has high spatial and temporal resolution and can carry abundant motion information, but it is difficult to maintain a long time effective record in practical application. In recent years, many studies have shown that the local field potential signal also contains abundant motion information, but its long-term stability is relatively less studied. In addition, the characterization and resolution of the local field potential signals in different frequency bands for the motion of the ipsilateral limbs are still unclear. In this study, we used silicon array electrode of Blackrock to record the neural signals in the primary motor cortex and the dorsal premotor area of the three macaques during the four-direction center-out experiment based on the brain computer interface platform of the rhesus monkey, and recorded the neural signals from the characteristics of the neural signals in the primary motor cortex and the dorsal premotor area of the rhesus monkeys. The long-term stability of the local field potential signal and the analysis of the ipsilateral motion information are analyzed and studied in three aspects: the correlation between the nerve signal and the motion signal and the decoding of the nerve signal. The results show that relative to the neuron spike signal, the recording quality of the local field potential signal attenuates slowly in the long term and can maintain the movement direction information for a longer time. When the recording quality of the neuron spike signal is poor, the local field potential signal can show better decoding performance. In particular, the local field potential signal can still provide decoding information when the whole neural electrode array can not record the neuron spike signal. In the aspect of ipsilateral motion representation and analysis, local field potential signals of multiple frequency bands represent the motion information of ipsilateral limbs. At the same time, the same neural signal showed a significantly different preference direction from the contralateral motion. The decoding results show that multiple frequency bands can accurately parse the motion information of the ipsilateral limbs, and the decoding performance of the 200-400Hz band is the best. In this study, the decoding performance of the local field potential signal with poor recording quality of the neuron spike signal is analyzed, and the characterization and analysis of the information of the local field potential signal on the ipsilateral limb motion in different frequency bands are deeply studied. The main innovations are as follows: (1) in the four-direction center-out motion paradigm, it is proved that the local field potential signal has better long-term stability than the neuron spike signal from three aspects: signal characteristics, directional tuning and off-line decoding performance; (2) when the recording quality of the neuron spike signal is seriously attenuated or even completely disappeared, the local field potential signal still has some physiological significance in the four-direction classification and decoding; (3) the results show that the local field potential signals in multiple frequency bands contain effective information about the motion of the ipsilateral limbs, especially in the 200-400Hz band, the decoding performance of the signals is relatively good, and the classification accuracy of the four directions is the highest. In this paper, the long-term stability of the local field potential signal and the characterization and analysis of the ipsilateral limb motion information show that the local field potential signal of the cerebral motor cortex also contains abundant limb motion information. It can be used as a stable neural signal source, and the related research results will further promote and enhance the clinical application of implanted brain-computer interface (BCI).
【学位授予单位】：浙江大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN911.7
本文编号：2200159
[Abstract]:Physical injury or nervous system diseases often lead to permanent loss of motor function, and the emergence of brain-computer interface technology brings new hope to these disabled patients. Brain-computer interface directly extracts motor information from the brain, bypasses the conventional spinal cord and peripheral nervous system, and directly controls the peripheral equipment, which provides a new idea for the reconstruction of motor function. At present, there are two kinds of nerve signals commonly used in implantable brain-computer interface: neuron spike signal and local field potential signal. The neuron spike signal has high spatial and temporal resolution and can carry abundant motion information, but it is difficult to maintain a long time effective record in practical application. In recent years, many studies have shown that the local field potential signal also contains abundant motion information, but its long-term stability is relatively less studied. In addition, the characterization and resolution of the local field potential signals in different frequency bands for the motion of the ipsilateral limbs are still unclear. In this study, we used silicon array electrode of Blackrock to record the neural signals in the primary motor cortex and the dorsal premotor area of the three macaques during the four-direction center-out experiment based on the brain computer interface platform of the rhesus monkey, and recorded the neural signals from the characteristics of the neural signals in the primary motor cortex and the dorsal premotor area of the rhesus monkeys. The long-term stability of the local field potential signal and the analysis of the ipsilateral motion information are analyzed and studied in three aspects: the correlation between the nerve signal and the motion signal and the decoding of the nerve signal. The results show that relative to the neuron spike signal, the recording quality of the local field potential signal attenuates slowly in the long term and can maintain the movement direction information for a longer time. When the recording quality of the neuron spike signal is poor, the local field potential signal can show better decoding performance. In particular, the local field potential signal can still provide decoding information when the whole neural electrode array can not record the neuron spike signal. In the aspect of ipsilateral motion representation and analysis, local field potential signals of multiple frequency bands represent the motion information of ipsilateral limbs. At the same time, the same neural signal showed a significantly different preference direction from the contralateral motion. The decoding results show that multiple frequency bands can accurately parse the motion information of the ipsilateral limbs, and the decoding performance of the 200-400Hz band is the best. In this study, the decoding performance of the local field potential signal with poor recording quality of the neuron spike signal is analyzed, and the characterization and analysis of the information of the local field potential signal on the ipsilateral limb motion in different frequency bands are deeply studied. The main innovations are as follows: (1) in the four-direction center-out motion paradigm, it is proved that the local field potential signal has better long-term stability than the neuron spike signal from three aspects: signal characteristics, directional tuning and off-line decoding performance; (2) when the recording quality of the neuron spike signal is seriously attenuated or even completely disappeared, the local field potential signal still has some physiological significance in the four-direction classification and decoding; (3) the results show that the local field potential signals in multiple frequency bands contain effective information about the motion of the ipsilateral limbs, especially in the 200-400Hz band, the decoding performance of the signals is relatively good, and the classification accuracy of the four directions is the highest. In this paper, the long-term stability of the local field potential signal and the characterization and analysis of the ipsilateral limb motion information show that the local field potential signal of the cerebral motor cortex also contains abundant limb motion information. It can be used as a stable neural signal source, and the related research results will further promote and enhance the clinical application of implanted brain-computer interface (BCI).
【学位授予单位】：浙江大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN911.7
【参考文献】
相关期刊论文 前1条
1 ;Development of an invasive brain machine interface with a monkey model[J];Chinese Science Bulletin;2012年16期
，本文编号：2200159
本文链接：https://www.wllwen.com/kejilunwen/wltx/2200159.html
