Passive Wireless ECoG Monitoring On Multiple Subjects

Abstract

In this work, a system level design for a wireless, multi-subject ElectroCorticoGram (ECoG) monitor- ing system for epilepsy research on mice is presented. It is a continuation of previous collaborations between the Erasmus Medical Centre (EMC) neuroscience department and the section Bioelectronics at Delft University of Technology. The system is realized by using discrete components and custom PCB development. The innovative focus has been put on the design of a flexible wireless link that uses backscattering, Frequency Division Multiple Access (FDMA) and an Software Defined Radio (SDR). The combination of a high bitrate (up to 320 kbit/s) backscattering link that uses digitally generated subcarriers and FDMA to support multiple concurrent measurements has not been presented in litera- ture before. Weight, size and power consumption have been identified as the most stringent limitations. The 915 MHz ISM band is used for communication, for which the potential GSM interference should be taken into account. The design of the system contains the RHD2132 IC by Intantech to function as Analog Front-End (AFE), a Cortex M0+ low-power microcontroller (MCU) and an RF switch and chip-scale antenna for backscattering communication. The MCU communicates with the AFE and generates the data packets and the subcarrier frequency. This subcarrier is modulated on the main 915 MHz carrier by means of on-off-keying. A prototype receiver has been designed to be able to evaluate the wireless link. Up to three wireless tags were tested simultaneously with different subcarrier frequencies. Provided that the power levels of these subcarriers do not differ more than 6 dB, the receiver can distinguish between the different tags and thus FDMA on a backscatter link has been demonstrated. Furthermore, it was proven that using multiple, orthogonally oriented antennas on a tag can increase the backscattered power by an order of magnitude while providing more immunity to the location and orientation of the tag. Power consumption has been measured for 2 scenarios. For a 500 Hz sampling rate scenario, the complete system consumes 14.5 mW. For a 20 kHz sampling rate scenario, the power consumption amounts to 20.3 mW. In-vivo measurements of the AFE have been performed, showing a potential improvement over the current measurement set-up at the EMC. This improvement is mostly found in the reduction of 50 Hz interference and the compactness of the set-up. Additionally it was found that a signal resolution of 8 bit can be enough for epileptic seizure detection, which means the requirement on the wireless link can be reduced.