Implantable micro-device for epilepsy seizure detection and subsequent treatment

Emerging implantable microdevices hold great promise for the treatment of patients with neurological conditions. These biomedical systems have been exploited as unconventional treatment for the conventionally untreatable patients. Recent progress in brain-machine-interface activities has led the researchers to analyze the intracerebral EEG (icEEG) recording in real-time and deliver subsequent treatments.;We present in this thesis a long-term safe and reliable low-power microsystem-based implantable device to perform efficient neural signal recording, seizure detection and subsequent treatment for epilepsy. The main contributions presented in this thesis are reported in five journal manuscripts, published or accepted for publication in IEEE Journals, and many others such as Low Power Electronics, and Emerging Technologies in Computing. The proposed microsystem includes a low-power integrated circuit (IC) intended for real-time epileptic seizure detection. This IC integrates a front-end preamplifier and epileptic seizure detector. The preamplifier is based on a new chopper stabilizer topology that reduces noise and power dissipation. The fabricated IC was tested using icEEG recordings from seven patients with drug-resistant epilepsy. The average seizure detection delay was 13.5 sec, well before the onset of clinical manifestations. The measured total power consumption of this chip is 51 µW.;A closed-loop neurostimulator (CLNS) is next introduced, which is dedicated to automatically detect seizure based on icEEG recordings from intracranial electrode contacts and provide an electrical stimulation feedback to the same contacts in order to disrupt these seizures. The seizure detector chip and a dedicated FPGA-based electrical stimulator were assembled together with common recording electrodes to complete the proposed prosthesis. This CLNS was validated offline using recording from ten patients with refractory epilepsy, and showed excellent performance for early detection of seizures and subsequent self-triggering electrical stimulation. Total power consumption of the CLNS is 16 mW. Alternatively, focal drug injection is the promising treatment for epilepsy. A responsive focal drug delivery system based on a new asynchronous seizure detector is also presented. The later system with data-dependent computation reduces up to 49% power consumption compared to the previous synchronous neurostimulator. The detector was validated using icEEG recordings of 7 patients who have previously undergone intracranial investigations for epilepsy surgery. The triggering of the drug delivery system was tested and a predefined seizure suppression dose was delivered ∼16 sec after electrographical seizure onsets.;In order to improve performances of the whole implantable device, a novel subdural electrode is presented for better icEEG recording quality and delineation of the epileptogenic zone during the presurgical evaluation. Moreover, the new subdural grid electrode has a new shape and several attributes to address some of the problems (e.g. brain infection, swelling and hemorrhage) encountered with currently used subdural electrodes. These electrodes are tested in vitro saline solution and in vivo experiments for 3 weeks, and signal-to-noise ratio is improved up to 6 dB compared to using commercial electrodes.