The Study Of Information Feedback By Intracortical Micro-stimulation And Neural Information Reduction For Invasive Brain Machine Interfaces

The Brain-Machine interface(BMI)establishes a direct information communication pathway between the brain and external devices.On one hand,the motor intent could be read out form the brain to control the external devices.On the other hand,the external environmental information could be written back into the brain.Therefore,the BMI has a promising prospect of application in the sensorimotor function rehabilitation for the disabled and the treatment of diseases in nerve injury.After years of development,many remarkable achievements have been obtained in the invasive BMI field.And now it is stepping into the stage of optimizing and upgrading.At present,the BMIs are mainly facing with the following problems:(1)The lack of study on the information feedback for closed-loop BMI;(2)The trend of building closed-loop BMI has put forward higher requirements for motor decoding.For the first problem,this paper studied how to use the intracortical micorstimulation(ICMS)as a new feedback way in BMI.We first studied the characteristics of information input by the ICMS.Then we built a closed-loop BMI system which used the ICMS to provide feedback information.At last we verified that the ICMS played an effective part in system feedback by controlled experiments.For the second problem,this paper studied how to improve the decoding efficiency by neural information reduction.By analyzing the correlation between single neuron and the experimental task,we could select a few important neurons out of all to do the decoding,which could ensure the decoding accuracy as well as reduce the computational load.By this way,we achieved the purpose of improving the efficiency of decoding.The major work and innovations of this paper are as follows:1)The effects on the cortical neural activity by external vibration stimulation and ICMS have been studied,which has provided an experimental basis for the ICMS as information input method in closed-loop BMI system.2)A method which used the ICMS as information feedback way in bi-directional BMI has been presented.By building a closed-loop BMI experimental platform on rodents,the results verified that the ICMS played an effective part in system feedback.3)A method which used the mutual information to quantitatively evaluate and select the task related neurons has been presented.By doing the correlation analysis between neurons and experimental tasks on reaching and grasping decoding experiment in non-human primates,the neural information reduction was realized,which provided and effective way for increasing the decoding efficiency.The research results in this paper may provide some guidance and reference for the building of a complete and efficient closed-loop Brain-Machine Interface in the future work.