Invasive Brain-Machine Interface

With the rapid development of studies in brain science and computer technology, the brain-machine interface technology has made considerable progress. A brain-machine interface is a communication system that does not depend on the brain's normal output pathways of peripheral nerves and muscles. The brain-machine interface can provide an access of people with sensory, motor, or other disabilities of neural function to communicate with the outside world.The brain-machine interface is divided into non-invasive brain-machine interface and invasive brain-machine interface. By the microelectrode array implanted in certain region of the brain, the invasive brain-machine interface can acquire the accurate neuronal activity of this region with the advantage of high signal quality, good time and space resolution. These neuronal signals are then decoded to command signals to control the outside devices, so patients equipped with the invasive brain-machine interface do not need training in advance. Compared with non-invasive brain-machine interface, the invasive brain-machine interface has more advantage and has become a major research direction.The study of invasive brain-machine interface includes two major subjects. One is the neuronal signal recording device; the other is algorithms used to analyze the neuronal signal data. Here, we work on technologies covering part of the two subjects. In this paper, we firstly introduce two approaches to acquire the extracellular neuronal signals. One approach is using electrophysiological techniques to record the real neuronal signals from a certain region in brain. We design and implement a light weight, miniature neuronal signal recording system which can be placed on the head of a small animal. In this paper, we present the design and implementation of the analogy front end in detail and simply introduce the ideal to finish the design of a backpack. With the recording system mentioned above, we have acquired the neuronal signal on the hippocampal region of a pigeon. The other approach is simulating the neuronal signals by analyzing the structural property of the real neuronal signals for the convenience of algorithms testing. We also propose a spike detection method based on the piecewise optimal morphological filter. The structure elements of the morphological filter are constructed with a Gaussian function, and optimized to approximate the wave shape of action potentials presented in the neuronal signals by a proposed criterion and optimal procedure. Morphological filter with the optimal structure elements can suppress the background noise effectively. An adaptive threshold is also proposed to detect action potentials in the output signals of the piecewise optimal morphological filter, which make our algorithm can be performed automatically. We test our proposed spike detection algorithm with both the real neuronal signals and simulated signals, and we also compare our method with two established spike detection algorithms. The results show that our method achieved higher hit rate and precision.After all, in this paper we make researches on the design and implementation of neuronal signal recording system, approaches of simulating neuronal signals and the spike detection algorithm and implement the recording of neuronal signals, background noise suppressing and accurate spike detection which have laid a solid foundation for the work of spike sorting and spike time series decoding.