Figure 1.
The head mounted module connects to the rodent's head stage. The base station module connects to a PC running USB audio software. Bidirectional communication allows transmission of neural data to the PC, and configuration data to the rat mounted module. The RoSco has one communication module which includes two transceivers that simultaneously operate in parallel to transmit the neural waveform through a single combiner and antenna. The base station module has diversity reception with two communication modules that simultaneously receive the same entire neural waveform in parallel. This provides redundancy and if data packets are missing from one stream the base station can still reconstruct the full neural waveform.
Figure 2.
The RoSco head mounted module shown on a Long Evans rat.
The head mounted module consists of three PCBs and a small battery. While appearing relatively large in this photo, the head mounted module is light weight. The blue wire is the antenna. The red LEDs are for motion tracking using an overhead camera system.
Figure 3.
Diagram of one signal amplification and conditioning channel.
The diagram shows how the reference, band pass filter and gain can be configured. Note the programmable gain amplifier (PGA) is a part of the microcontroller saving a large number of external components and electronic complexity.
Figure 4.
Outdoor rodent test enclosure located in Brisbane, Australia.
The enclosure is 7×5 m, has a translucent roof and wire mesh walls. A rodent wearing RoSco can be seen towards the top of the arena.
Figure 5.
Measured transfer functions of the programmable bandpass filter.
The filter can be configured for both a narrow-band mode for AP recording (solid line) and a wide-band mode for LFP recording (broken line). The output of this filter stage is further amplified by a programmable gain amplifier.
Figure 6.
Measured common-mode rejection ratio.
The CMRR measured for a representative channel and the selected reference. The rejection ratio is high over most of the band of interest.
Figure 7.
Measured noise power spectral density.
This figure shows the measured noise power spectral density with the programmable band-pass filter programmed with the wide pass-band.
Figure 8.
Measured horizontal plane radiation profile of the rat head-mounted module (vertical polarization).
Figure 9.
Wireless transmission error rate.
Box plot of the error rate versus the distance between the rat module and the midpoint between the antennas. The figure shows that the practical limit for this wireless system with the given antenna configuration is 10
Figure 10.
Wireless transmission error rate with diversity reception.
Box plot of the error rate at 10
Figure 11.
Sample neural waveforms from the RoSco and Axona system.
100(top three traces, blue) and Axona (bottom three traces, red). The RoSco signal has been digitally filtered with a high-order high-pass filter at 300 Hz. The increased detail of the Axona signal is due to the difference in sample rate: RoSco at 20 kHz and Axona at 48 kHz. Scale bars at bottom right are 10 ms and 50 µV for x-axis and y-axis respectively.
Figure 12.
Comparison of characteristic unit spikes from the RoSco and Axona systems.
(a, b) Waveforms were isolated from two units (blue – unit 1; red – unit 2) using (a) RoSco (RMS SNR range 2.0–3.1) and (b) Axona (RMS SNR range 1.8–2.8). Scale bars at top right are 500 µs and 50 µV for x-axis and y-axis respectively. (c) Example dimensions from the unit clustering, for the two units in a and b, obtained using the WaveClus package for the RoSco (top) and Axona (bottom) data. Both dimensions are unitless feature space.
Figure 13.
Electrical recording and identified spike times from two units.
One unit from the dentate gyrus is shown with red stars in: (a) the raw trace and (b) a theta-filtered 4-12 Hz LFP. Unit activity appears more often on particular phases of the theta cycle, as expected from many earlier hippocampal studies [19], [20]. Scale bars at bottom left are 100 ms for x-axis and 50 µV/250 µV for top/bottom y-axes.
Figure 14.
Characteristic unit spikes captured by RoSco in the outdoor enclosure.
The unit spikes were taken from three wires on one tetrode. (a) The characteristic spike waveforms of two units. SNRs (RMS) are between 1.3 and 2.8 depending on the unit and the wire. Scale bars in top right are 500 µs and 50 µV for x-axis and y-axis respectively. (b) Example dimensions from the unit clustering obtained using the WaveClus package. Both dimensions are unitless feature space.
Table 1.
Comparison between RoSco and existing systems.