Study On Processing And Analysis Methods Of Cerebral Blood Flow Biological Impedance Signal

Brain function detection is the basis of brain science research.At present,brain function detection methods are developing in the direction of safety,efficiency and continuity.According to the regulation of blood flow by nerves in brain functional areas,the characteristics of blood circulation are of great significance to judge the state of brain functional areas.Bioimpedance detection technology meets the development requirements of brain function detection.According to the different impedance characteristics of cerebral blood flow in different states of the brain,impedance detection technology can be used to measure the impedance changes associated with cerebral blood flow,and then the neural excitation state of brain functional area can be judged by analyzing the impedance characteristics.Therefore,this paper studies the changes of hemodynamic parameters of cerebral blood flow impedance signal under brain functional activity and its automatic identification and classification approach,which is of vital significance to the evaluation of brain neural activity.Firstly,according to the generation mechanism of cerebral blood flow impedance signal,a visual stimulation experiment is designed to induce neural activity in the functional area of cerebral occipital lobe,and the MP160 multi-channel physiological recorder is needed to complete the signal acquisition.The collected cerebral blood flow impedance signal data in two different physiological states before and after stimulation are used to analyze the changes of hemodynamic parameters and build the database required for classification model.Aiming at the problem of noise interference in cerebral blood flow impedance signal,a signal processing method based on wavelet heuristic soft threshold denoising combined with wavelet decomposition and reconstruction to remove baseline drift is proposed,which can restore the effective signal well.Secondly,in order to verify whether the cerebral hemodynamic parameters can effectively reflect the differences of cerebral blood flow impedance signals in different conditions,and to study the changes of blood flow and vascular state in the process of cerebral nerve activity simultaneously.the hemodynamic parameters are calculated and analyzed respectively for the simulation data before and after the reduction of vascular elasticity and the measured two kinds of data before and after visual stimulation.The results indicate that the cerebral hemodynamic parameters could characterize the state of blood vessels and blood flow in the brain.The local blood flow in the cerebral occipital lobe functional area increases and the vascular elasticity increases relatively before and after visual stimulation.The results not only verify that the external intervention measures can effectively change the mechanism of cerebral blood flow regulation,but also prove the feasibility and effectiveness of the stimulation experiment and detection scheme designed in this paper.Finally,combining the commonly used data processing methods with deep learning algorithm,a cerebral blood flow impedance signal classification approach based on convolutional neural network（CNN）is proposed.Taking advantage of the characteristics that CNN can automatically extract the inherent law and representation level of input data through training,the periodic segments of cerebral blood flow impedance signals in two physiological states are directly input into the CNN built in this paper to complete the signal recognition and classification,and the final classing accuracy is 86.9%.The result indicates that the approach presented in the paper is suitable for the classification of cerebral blood flow impedance signals.