文摘
Electrical impedance tomography EIT) is a non-invasive technique for imaging changes in the electrical conductivity of tissues within a body section from impedance measurements performed with surface electrodes. The Institut de genie biomedical IGB) at Ecole Polytechnique de Montreal has been involved in the development of this technique since 1987. Recent projects of the IGB include developing a system for monitoring epileptic patients that are candidates for surgery. This system is designed to acquire EIT and electroencephalographic EEG) data simultaneously. The rationale for combining EIT and EEG data is to improve the accuracy with which areas of the brain responsible for triggering epileptic seizures can be localized. Accurate localization of these foci allows,in many cases,to cure the patient from epilepsy by surgically removing these areas of the brain. Although various clinical tests and medical imaging procedures are presently used to localize epileptic foci,confirmation of their positions often require a direct electrophysiological investigation with implanted electrodes. This is a highly invasive procedure that carries a significant risk of hemorrhage. The system we are developing could eventually replace direct investigations with intracranial electrodes by a simple noninvasive imaging technique that can be deployed at relatively low cost. The combined EIT-EEG data acquisition system has four components: 1) an array of 24 scalp electrodes,2) a scan head module containing the front-end circuits required to measure the impedance of cerebral tissues and record EEG signals,3) a base station based on a field programmable gate array FPGA) that performs all digital signal processing and system control operations,and 4) a computer that receives the data from the base station and reconstructs images of brain conductivity changes. The primary goal of this masters project was to develop the base station module. This was achieved by performing the following tasks. 1) We developed the configuration code for the FPGA of a base station module previously built by the IGB. The code was written in VHDL,a high-level language,in order to facilitate maintenance and portability to future systems. 2) The base station module was tested at the board and system level by measuring three performance indicators: a) spurious-free dynamic range SFDR),b) signal to noise ratio SNR),c) overall accuracy,and d) thermal drift. 3) The performance indicators measured on the new system were compared to those of an older system that uses discrete specialized chips for signal processing. 4) Preliminary in vivo tests were done with the new system. SFDR was measured with a network/spectrum analyzer,while SNR and overall accuracy were evaluated with functions built in the user interface. SNR and overal accuracy were obtained from data acquired on a precision resistor mesh phantom that includes circuits to emulate electrode-skin contact impedances. The phantom was connected to the scan head module with the same electrode leads used for in vivo data acquisition. This setup provides reproducible test conditions while accounting for the interference and noise sources encountered in clinical recording environments. Comparisons with the older system were done with data acquired at 5 frames/s,using an applied current of 2 mA peak at 50 kHz. Results for the system comparisons showed improvements in all performance indicators. The average SFDR for the 20 to 65 kHz frequency range increased from 52.1 dBm for the older system to 57.7 dBm for the new system. The average SNR increased from 61.5 dB to 64.5 dB,and overall accuracy from 99.8% to 99.9%. Thermal drift was about the same in the two systems. Preliminary in vivo tests were done with 24 electrodes surrounding the torso at the level of the 8th intercostal space. These tests showed that images of lung ventilation could be reconstructed in real time at up to 20 frames/s. We were able to record electrocardiograms ECG) at a sampling rate of 1,3 kHz per channel from the same 24 electrodes by reducing the voltage gain of the EEG amplifiers. However,with electrodes placed around the head we were unable to obtain coherent EIT images. Also,at high gain,signals from the EEG amplifiers were dominated by interference from the power mains. Causes for failure to image brain conductivity changes and to record EEG have been identified. Part of the problem lies in having omitted in the scan head design to use shielded electrode leads with shield drivers. This choice was,in retrospect,wrongly justified by ergonomic considerations. A revision of the scan head module will be required to correct this problem. For the base station module,we describe in the final chapter of the dissertation five projects aimed at improving the SNR of the system. In conclusion,we have contributed with this project in developing a non-invasive technique for imaging changes in brain blood flow and volume,while monitoring brain electrical activity. The application of this technique to the localization of epileptic foci is a long-term goal. It will require further advances in system hardware and image reconstruction algorithms,as well as a pilot study on patients admitted to the epilepsy unit for a presurgical evaluation. If this validation is successful,combined EIT and EEG monitoring could be at the fore-front of non-invasive presurgical tools,due to its safety,low-cost,and high temporal resolution.