The Neurochip-2: an autonomous head-fixed computer for recording and stimulating in freely behaving monkeys

Abstract

The Neurochip-2 is a second generation, battery-powered device for neural recording and stimulating that is small enough to be carried in a chamber on a monkey's head. It has three recording channels, with user-adjustable gains, filters, and sampling rates, that can be optimized for recording single unit activity, local field potentials, electrocorticography, electromyography, arm acceleration, etc. Recorded data are stored on a removable, flash memory card. The Neurochip-2 also has three separate stimulation channels. Two "programmable-system-on-chips" (PSoCs) control the data acquisition and stimulus output. The PSoCs permit flexible real-time processing of the recorded data, such as digital filtering and time-amplitude window discrimination. The PSoCs can be programmed to deliver stimulation contingent on neural events or deliver preprogrammed stimuli. Access pins to the microcontroller are also available to connect external devices, such as accelerometers. The Neurochip-2 can record and stimulate autonomously for up to several days in freely behaving monkeys, enabling a wide range of novel neurophysiological and neuroengineering experiments.

Fig. 1

(A) Neurochip-2 inside the titanium casing (attached via nylon screws to a polycarbonate base, not visible). The battery is housed in the polycarbonate lid, shown to the right. (B) The two versions of the Neurochip-2: a compact, lower output current version (left) and a larger, high output current version (right). The gold access pins (bottom right of boards) allow connections to external devices and access four data channels on PSoC-B.

Fig. 2

(A) Block diagram of the major Neurochip-2 components and signal routing. G: gain, LF: low-pass filter cutoff (Hz), PSoC: programmable system-on-chip, SPI: serial peripheral interface, AVR: Atmel AVR microcontroller, DAC: digital-to-analog converter. Connections between the battery and active components are not shown. (B) Circuit diagram of one of the “V to I converter” blocks.

Fig. 3

Matlab-based graphical user interface on a PC. Upper bank of windows shows uploadable parameters for recording, spike discrimination and stimulation. Middle window illustrates action potentials recorded on channel A (red traces) and discriminated via a threshold (horizontal line) and two time-amplitude windows. Window below shows 3 s of raw data from three intracortical electrodes: neuronal activity on channel A (black trace), and wide-band (10–500 Hz) LFPs on channels B and C (green and yellow traces). Blue lines indicate spike times detected by window discriminator. Lowest window shows binned activity over 10-s interval.

Fig. 4

Bench testing results. (A) Measured frequency response for three different filter settings of channel A operating at maximum sampling rate (24 kS/s). (B) Measured output of the stimulator for two different intensities and pulse widths.

Fig. 5

Thirty-five-hour continuous ECoG recording from a cortical surface electrode over the precentral gyrus. The upper plot shows a spectrogram of the recording (power calibration at right). Three short segments of raw data are shown in the lower plots, corresponding to times indicated by the dots above the spectrogram.

Fig. 6

Eighteen-hour continuous recording of an intracortical local field potential from left motor cortex (top) and right forearm acceleration (bottom). The LFP spectrogram (upper plot) and average acceleration (bottom plot, in digitizer units) were averaged over 30-s bins.

Fig. 7

EMG responses evoked by stimulation of cortical surface using Neurochip-2 (version HCV). Single, biphasic stimuli of 2 mA intensity were delivered every 20 s. Bipolar EMG was continuously recorded from the contralateral flexor carpi radialis muscle. Shown are the average EMG responses during consecutive 4-h-long periods, for stimuli that occurred in the presence of background EMG activity (hence the fewer triggers during night time).

Fig. 8

Closed-loop operation of Neurochip-2. Action potentials were recorded from one intracortical electrode on channel A (500 Hz highpass filter; blue trace) and field potentials were recorded from two other intracortical electrodes on channels B and C (10 Hz highpass filter). The Neurochip was programmed to deliver a 70-μA, 0.2-ms biphasic stimulus pulse to a fourth intracortical electrode 5 ms after each discriminated spike. The experiment ran continuously for 10 h. The top plot shows a short segment of raw data, with two spike-stimulus pairs. The bottom plot shows the average spike waveform and stimulus artifact on channel A. Amplitude in both plots is shown over the full range of digitizer units (adu).