Hardware for seizure prediction: towards wearable devices to support epileptic people

This paper presents the hardware developed for the EPILEPSIAE project (www.epilepsiae.eu), focused on epileptic seizure prediction. A portable low power acquisition system for EEG signals, called LTM-EU (Long Term Monitoring), with 64 channels and 2048 Hz sampling rate each and a safe (high isolation) PC interface on a PCIe bus specifically designed for this task, is described. The acquisition system, designed for a rapid commercialization, though used for research purposes, got the CE certification. The signal from the patient, on each channel, is amplified, converted in digital form and stored into a local flash memory (SD-MMC, 4 GB). Data are then formatted into a serial stream at 4 Mb/s and sent through a half-duplex RS485 link to the host where a specifically designed PCIe (BQPCIe) interface receive them and release the information to the OS (Windows or Linux). The amplifier runs with a couple of AA battery for more than 15 hours (300 mW). If a wireless link is established (Bluetooth), a bandwidth limited stream of data (or a subset of channels) is sent for monitoring purposes. The mission is to support the researchers of the consortium with a suitable hardware to have a real time seizure prediction system for algorithms tests. In the experimental phase all algorithms run on a portable PC, wire or wireless connected to the acquisition system.