Design Of Brain Electrical Impedance Tomography Detection Platform

In the field of Brain medicine,Cerebral hemorrhage is a common complication in many diseases,such as cerebral stroke(Cerebral Vascular Accident,CVA)and Traumatic Brain Injury(Traumatic Brain Injury,TBI).For intracerebral hemorrhage,the critical treatment time in the early stage is short and very valuable.Therefore,rapid and accurate detection of cerebral hemorrhage and correct and timely treatment are very important for the recovery of patients without aggravating the condition of cerebral hemorrhage and causing no side effects.Electrical Impedance Tomography(Electrical Impedance Tomography,EIT)is a kind of technology for detecting the pathological changes through imaging.Its non-invasive and harmless features make it very suitable for the detection of brain diseases.Based on the theoretical basis of bioelectrical impedance imaging technology and electrical characteristics of brain tissue,this paper designed a brain electrical impedance imaging detection platform,and studied the method of detecting cerebral hemorrhage combined with EIT and machine learning.First of all,due to the obstruction effect of high resistance cranium on excitation current and the shunt effect of cerebrospinal fluid on excitation current,it is difficult to obtain intracranial impedance changes.A programmable excitation source is developed to compensate the excitation current,including: A programmable signal generator is designed using DDS+DAC technology to output the amplitude-modulating sine wave signal with frequency range of 10 k Hz to 1MHz.A pseudo-short circuit is added to the mirror voltage controlled constant current source to reduce the shunt effect caused by distributed capacitance at the electrode interface,and the input current at the return end of the excitation source is collected to compensate the output excitation current.To increase the excitation current into the intracranial;Aiming at the high common-mode voltage between electrodes,a differential voltage detection circuit is developed to improve the common-mode rejection ratio,which can be kept above 60 d B in the frequency band.Secondly,considering the real-time performance of impedance information demodulation system and the noise suppression ability of demodulation algorithm,the multi-frequency digital orthogonal sequence demodulation algorithm is adopted,which not only guarantees the real-time performance of the system,but also effectively inhibits the random noise of the system.In view of the current feasibility of static algorithm imaging in the detection of cerebral hemorrhage and the difficulty in obtaining reference data of general cerebral patients before the onset of the disease,FEM numerical model in the brain was established,and the radial basis kernel-based support vector machine(SVM)was directly combined with the measurement frame of EIT to classify and detect the size and location of the blood clots in cerebral hemorrhage.By simulating various hemorrhagic lesions and conducting test classification experiments,the results showed that the accuracy of detecting the location of bleeding clots could reach 90%,and the accuracy of detecting the size of clots was also 80.833%.Finally,experiments are designed to test the load capacity of excitation source output and the accuracy of acquisition circuit signal acquisition,and the feasibility of the system is verified by physical model imaging experiments and human body imaging experiments.The experimental results show that the maximum difference of excitation current output of excitation source is about 0.5% when it ranges from 100 Hz to 1MHz.The average SNR of signal acquisition is kept above 79 d B,and the disturbance impedance changes and disturbance positions of the target can be accurately reflected by the physical model and human experimental imaging results,which verifies the feasibility of the platform.