Large-scale electrophysiology recordings are recognized as a powerful tool for systems neurobiology and investigation of normal and pathological brain function (Buzsaki, 2004). Continuous, high, spatial and temporal resolution intracranial electroencephalography (iEEG) and single-neuron recordings from animals and humans are increasingly being used to investigate both normal and pathologic brain functions, (e.g. Kraskov et al., 2007; Gelbard-Sagiv et al., 2008). There is accumulating evidence that the bandwidth used for clinical iEEG is inadequate, and that high frequency oscillations can localize epileptogenic brain (Gardner et al., 2007; Bragin et al., 2002; Urrestarazu et al., 2007; Worrell et al., 2004, 2008). These opportunities have not been fully exploited, however, due to limitations in data transmission and storage technologies that limit data acquisition directly by reducing the duration, number of channels, sampling rate or resolution of recordings in order to generate manageable amounts of data. Performance of devices that acquire and store or transmit iEEG data would benefit significantly from efficient on-board data compression. We propose here to implement the Range Encoded Differences (RED) algorithm (MEF REF), an open source established compression strategy for electrophysiologic data, on our Cube 2, data acquisition device. The Cube 2 acquires up to 128 channels of high bandwidth iEEG and transmits it via WiFi, or stores it locally to an SD card. The device has a powerful Zynq 7000 series (Xilinx, inc.) Field Programmable Gated Array (FPGA) running a Linux operating system. Implementation of this data compression will reduce transmission and storage requirements to 3-4 bits per sample in preliminary testing, a doubling of it's current capacity. For Phase 1 of this proposal we are focusing on implementation in existing hardware. If successful it will represent a significant product improvement which could be made commercially available almost immediately. Going forward, however, we hope to move the compression into an application specific integrated circuit (ASIC) to reduce power consumption in size in a phase 2 application for this work. Such an chip could have potential application in any device that acquired and stored or transmitted iEEG data, such as next generation human implantable devices (e.g. NeuroPace or Medtronic RC+S) or any continuously sampled time series data such as ECG. The ASIC would be made commercially available independent of the Cube 2 system.
Public Health Relevance Statement: High bandwidth neural data is becoming increasingly important in conducting neurophysiological research. The work in this proposal is aimed at developing embedded compression of EEG on a state of the art wireless data acquisition system. The successful completion of this project will lest substantially increase the data storage and transmission capacity of this system, and be applicable to a multitude of systems in the future.
Project Terms: Algorithms; Animals; Brain; Chronic; Clinical; Code; Consumption; Data; data acquisition; Data Compression; data exchange; Data Storage and Retrieval; Devices; Electrocardiogram; Electroencephalography; Electrophysiology (science); Epilepsy; Evaluation; Future; High Frequency Oscillation; Human; Implant; implantable device; Investigation; Life; Linux; Neurobiology; Neurons; neurophysiology; next generation; open source; Operating System; Pathologic; Performance; Phase; programs; relating to nervous system; Research; Resolution; Rodent; Running; Sampling; Series; Source Code; Speed; System; Technology; temporal measurement; Testing; Time; tool; transmission process; Wireless Technology; Work
