A fully reconfigurable controller dedicated to implantable recording devices

Abstract

Neural signals monitoring systems become a key issue to cortical bioelectrical comprehension research. This paper describes the implementation of a dynamically reconfigurable controller dedicated for real-time implantable acquisition systems. The authors presented the cortical signal's characteristics and the requirements needed to extract and reconstruct the signal once transmitted outside the implant. The implementation of a wavelet transformation used as a feature extraction for spike detection, compression and denoising techniques were reported. A packet based protocol permitted to communicate in both directions with the implant trough the wireless link. A complete 32 channels prototype based on an optimized pipelined architecture was implemented and the resources' usage for further expansion was discussed.