ABSTRACT:

In this paper, we present a single voltage-window count-based seizure onset detection algorithm and its associated hardware implementation. The proposed algorithm combines the advantages associated with voltage-window count-based and event-based threshold-voltage detections. The result is an algorithm that is more tolerant to noise, dc offsets, baseline energy variations, and seconds-long nonseizure related sharp activities. In addition, only one parameter (one threshold voltage) needs to be optimized per patient, and for that, only one seizure per patient is used for training, making the process of optimizing the patient-specific detector a simple task. The time evaluation period when counting is performed is kept constant across all patients studied, and is fixed at 5 s in this work. A novel dual path digital signal processing unit in the back-end of the detector is included and is shown to decrease the detection latency by 14%. Experimental results on a printed circuit board using commercially available discrete components confirm the correct functionality of the proposed detector. The proposed algorithm achieves 100% sensitivity, 10.7 s average detection delay, and a single false alarm when evaluated on a total of 25 seizures and 24 nonseizure datasets of intracerebral electroencephalographic (icEEG) recordings from five patients from the epilepsy monitoring unit of Notre-Dame Hospital in Montréal. In addition, monolithic integration of the overall system, including bio-amplification and comparison, is also carried out in a TSMC 0.18-$mu$m complementary metal–oxide–semiconductor technology. Simulations show that a static power dissipation of 7 $mu {rm W}$, 99% of which is consumed by the front-end bio-amplifier, is achieved, showing the potential of using the proposed seizure detector in a closed-loop b-
ain–prosthesis device interface for seizure control and treatment.