Transformee en ondelettes pour un systeme d'acquisition de signaux corticaux implantable (French text).

摘要

Recent advances in electronics and microfabrication have led to the development of novel brain-computer interfaces composed of microelectrode arrays combined to integrated circuits. These new biomedical devices could enable people suffering from neurological dysfunctions to control prostheses or to recover sensations by using a cortical implant composed of microelectrodes arrays that would record or simulate extracellular biopotentials. Currently, projects involving multichannel neuronal signal acquisition systems are undertaken by several research teams and some of them are already able to measure cortex reaction to visual stimuli and to control robotized devices with basic movements.; Implementation of continuous wavelet transform with analog filters has been preferred to discrete transform because it is less costly in power and integration area, which are two unavoidable requirements for implantable circuits.; Log-domain principle has been used for filter implementation to take advantage of the very low power consumption inherent to weakly inverted CMOS transistors. The state-space model has been balanced so all its parameters are in the same range for a greater precision of frequency and time responses. (Abstract shortened by UMI.)