Autonomous head-mounted electrophysiology systems for freely behaving primates

Abstract

Recent technological advances have led to new light-weight battery-operated systems for electrophysiology. Such systems are head mounted, run for days without experimenter intervention, and can record and stimulate from single or multiple electrodes implanted in a freely behaving primate. Here we discuss existing systems, studies that use them, and how they can augment traditional, physically restrained, 'in-rig' electrophysiology. With existing technical capabilities, these systems can acquire multiple signal classes, such as spikes, local field potential, and electromyography signals, and can stimulate based on real-time processing of recorded signals. Moving forward, this class of technologies, along with advances in neural signal processing and behavioral monitoring, have the potential to dramatically expand the scope and scale of electrophysiological studies.

Figure 1

Autonomous head-mounted electrophysiology systems overview: (A) shows a system diagram summarizing potential autonomous head-mounted electrophysiology system designs. (B) shows a sketch of how a wireless telemetry recording system operates and provides a general sense of scale of these systems. This is the recording setup used in HermesC [49] studies, in which one broadband channels and multiple threshold channels are recorded and synchronized to recorded video.

Figure 2

The Hermes system and single unit stability across days. (A) Shows the 95% confidence intervals of sorted action potentials from a single channel recorded with HermesB [24] accross 29.5 hours, note that the action potentials of the units are flexing relative to one another. (B) Shows a large shift in peak-to-peak waveform amplitude coincident with a large accelerometer event. (C) Is a drawing of the HermesC system [49] from implant to electronics. (D) is two pictures of the actual system: on the left is the chassis base, Utah array connector, and board stack; on the right is the fully enclosed system with transmitting antenna. (E) shows the 95% peak-to-peak envelope of waveforms observed for 12.5 days for channels 2-20 with HermesC. The voltage is averaged across 3 hour bins, normalized to the average voltage on that channel. Variations fell within 65% - 215% of the mean, or 9.3 uV - 220 uV. As shown in (F), channels can remain static; although single units often appear and disappear.

Figure 3

The Neurochip system and multi-day stimulation induced plasticity. (A) is a drawing of the neurochip system [48] and (B) is a picture of the neurochip board. (C) is a schematic of the paradigm employed by Jackson et. al. in [22], in which an electrode with a sortable unit, Nstim, was stimulated after a delay from when Nrec, a unit recorded on another electrode, spiked. The paradigm was run continuously for 2 days with the neurochip. After the two days, tuning curves for Nstim, Nrec, and a control unit were recorded with a wrist flexion task. (D) shows the tuning curves plotted relative to the axis labels shown. Note that the tuning of Nstim after stimulation is more similar to Nrec than before stimulation and that the tuning of the control cell is relatively unchanged.

Figure 4

Expanding the scope of motor system studies: (A) shows a progression of potential motor control studies enabled by head-mounted electrophysiology systems. 1 is the traditional wired in-rig setup in which the animal is engaged in a reaching task with a highly constrained posture. 2 utilizes a wireless telemetry system and the animal is able to move to many different postures as he plays with a sensorized manipulandam. 3 shows the possibility of studying complex movements, such as leaping. 4 is the ability to decode motor intentions. (B) is a sketch of a potential computer vision system for recording detailed body posture simultaneously with wireless neural data. The images from all cameras are used to construct a model of all joint positions at every moment in time.

Figure 5

An example of in cage single trial analysis and decoding: (A) shows the class of behavior marked in the traces below. Monkey D was engaged in a free-pace/free-posture reaching task to acquire treats placed on a platform in front of her cage. (B) shows the first two orthogonal neural dimensions found by smooth and factor analysis [90] during four minutes of intermittent reaching. Projections were calculated from 14 simultaneously recorded threshold channels collected with HermesC [49]. Color coding marks the period of time for outward and inward reaches consistent with the color of the arrows in (A)