Medical & Biological Engineering & Computing

, Volume 46, Issue 8, pp 833–839

A low noise remotely controllable wireless telemetry system for single-unit recording in rats navigating in a vertical maze

Authors

  • Hsin-Yung Chen
    • Institute of Biomedical EngineeringNational Cheng Kung University
  • Jin-Shang Wu
    • Department of Family Medicine, Medical CollegeNational Cheng Kung University
  • Brian Hyland
    • Department of PhysiologyUniversity of Otago
  • Xiao-Dong Lu
    • Department of PhysiologyUniversity of Otago
    • Institute of Biomedical EngineeringNational Cheng Kung University
Technical Note

DOI: 10.1007/s11517-008-0355-6

Cite this article as:
Chen, H., Wu, J., Hyland, B. et al. Med Biol Eng Comput (2008) 46: 833. doi:10.1007/s11517-008-0355-6

Abstract

The use of cables for recording neural activity limits the scope of behavioral tests used in conscious free-moving animals. Particularly, cable attachments make it impossible to record in three-dimensional (3D) mazes where levels are vertically stacked or in enclosed spaces. Such environments are of particular interest in investigations of hippocampal place cells, in which neural activity is correlated with spatial position in the environment. We developed a flexible miniaturized Bluetooth-based wireless data acquisition system. The wireless module included an 8-channel analogue front end, digital controller, and Bluetooth transceiver mounted on a backpack. Our bidirectional wireless design allowed all data channels to be previewed at 1 kHz sample rate, and one channel, selected by remote control, to be sampled at 10 kHz. Extracellular recordings of neuronal activity are highly susceptible to ambient electrical noise due to the high electrode impedance. Through careful design of appropriate shielding and hardware configuration to avoid ground loops, mains power and Bluetooth hopping frequency noise were reduced sufficiently to yield signal quality comparable to those recorded by wired systems. With this system we were able to obtain single-unit recordings of hippocampal place cells in rats running an enclosed vertical maze, over a range of 5 m.

Keywords

Single-unit recordingWireless telemetryFreely moving ratsPlace cell

1 Introduction

Multichannel neuronal recordings using multi-wire electrodes attached to measurement systems with cables are commonly used to record from brain structures during behavior. The hippocampal place cells firing at higher rates when animals enter a particular location are involved in learning and using spatial cues to solve spatial problems by [2, 4, 9, 10, 13]. However, progress in this area has been limited by the requirement that animals must be restrained and tethered by the cable recording system, restricting studies to two-dimensional (2D) open field arenas or mazes [6, 7, 12]. Cable systems make it particularly difficult to record in environments that include enclosed spaces such as doorways or tunnels, or that allow animals to navigate in vertically arranged maze arms.

Wireless biopotential recording modules provide an elegant alternative to replace the cabled systems to record in enclosed three-dimensional (3D) spaces [6, 12]. Bluetooth wireless communication is characterized by low power demand, and light-weight making it well suited for recordings from small animals. Furthermore, Bluetooth can be configured for bi-directional communication. Current applications of Bluetooth-based wireless transmission have focused on implantable wireless modules for cardiovascular signals or sympathetic nerve activity [1, 14]; however, these types of recordings have less stringent requirements for sampling rate or source impedance compared to brain single-unit recording.

The aim of this study was to implement a Bluetooth-based wireless transceiver module with remotely controlled selection of recording channel which could meet the requirements for single-unit recording of small animals. Design requirements were that the unit should be small and light enough to be carried by a rat, have adequate sampling rate for neural recording, and be robust against electrical interference. We tested the feasibility and reliability of our system, by recording neural activity from hippocampal area CA1 of rats navigating in a vertical maze.

2 Methods

The Bluetooth-based telemetry system we developed consists of two separate components: a wireless neural recording module in a slave mode and a PC-host telemetry control subsystem, as depicted in Fig. 1 [3].
https://static-content.springer.com/image/art%3A10.1007%2Fs11517-008-0355-6/MediaObjects/11517_2008_355_Fig1_HTML.gif
Fig. 1

Block diagram of Bluetooth-based single-unit recording system. The wireless data acquisition module is carried as a backpack to the rat and transmits to PC-based control subsystem. Microdrive wire electrodes were implanted to CA1 of hippocampus for sensing single-unit activity which was amplified by analogue front end amplifier and digitized by A/D connecting to MCU for Bluetooth wireless transmission to the host PC

2.1 Wireless data acquisition module for single-unit recording

The wireless data acquisition module, carried as a backpack on the rat, amplifies and samples the brain neural activity for wireless transmission. The analogue front end (AFE) consists of miniature field-effect transistors (FET), 8-channel of buffer amplifiers, preamplifiers and second-stage amplifiers. Miniature N-channel FET which possesses high-input impedance (~1012 Ω) were placed directly on the headstage to match the impedance of microdrive wired electrodes. The AFE incorporates 8 J-FET unity-gain voltage followers as input first-stage amplifiers (TL064C, Texas Instruments, USA). After the voltage follower, the signals are amplified by two-stage amplifiers—a precision instrumentation amplifier with a gain of 25 (INA2128, Burr-Brown, USA) offering high common mode rejection ratio (CMRR) of 120 dB minimum and a bandpass second-stage amplifier with a gain of 100—to achieve a gain of 2,500 at a passing bandwidth between 5 and 3.0 kHz. Both the single-unit recording of place cells and theta-band brain wave (5–10 Hz) can be recorded for spatial learning study in our design.

The microcontroller unit (MCU, PIC18F452, Microchip, USA) is the core of our Bluetooth-based neural recording module, which connects to the 12-bit A/D converter (MCP3206, Microchip) for sampling single-unit activities after amplification. The external commands sent from Bluetooth host PC are decoded by the MCU to set up the data acquisition protocols, including sampling channel and sampling rate. After data sampling, the MCU configures two bytes of encoded data in universal asynchronous receiver-transmitter (UART) format, then adding 4-bit channel information as a time marker to enable checking for data loss during wireless transmission.

The Bluetooth module, BT-20 (RainSun), is small (2.5 mm × 14.5 mm) and light-weight (0.4 g) and provides a nominal range of up to 10 m (at 0 dBm). The device is compliant with power class 2 of Bluetooth version 1.1 and has a data rate of 750 kbps. The Bluetooth version 1.1 technology currently operates in the unlicensed 2.4 GHz industrial, scientific and medicine (ISM) frequency bands and utilizes frequency hopping spread spectrum (FHSS) at 1,600 hops/s to protect against transmission disturbance.

The whole telemetry system is powered from a pair of 170 mAh at 3.7 V rechargeable lithium ion batteries (AHB402030, Synergy). Separating the power sources of the Bluetooth module and the highly sensitive AFE minimized RF interference from the Bluetooth unit avoided a ground loop, which can generate white noise interference. Shunt capacitors to the power source especially for the front end stages and appropriate shielding of the entire AFE further reduced the coupling of power line ripples. The rechargeable battery allows about 120 min of continuous operation which is sufficient for about 60 min of validation experiments.

2.2 PC-based telemetry control subsystem

The PC-based control system was designed to utilize the built-in functions of the Bluetooth development tool, BlueSuite (CSR, UK), for downloading the firmware, source codes, and transmission parameters to the Bluetooth module. To achieve higher transmission rate through the RS232 interface, a high speed RS-232 Card (PCI-1610A, Adventech) was used in this study which provides up to 460.4 kbps. This setup enables our system to transmit 8-channel recording simultaneously at 1 kHz sampling rate per channel for previewing, and recording of a single channel at 10 kHz sampling rate. Upon receiving an external command, the MCU sets up the number of channels to record and their sampling rate before the interrupt service for A/D converter and the acquired data are transmitted to the computer.

A graphical user interface (GUI) developed using LabVIEW (National Instrumentation, USA) was used to control data acquisition and delivery of external commands, to preview the signal and record the data for off-line analysis. Standard signal unit analyses were performed using custom-built routines in Matlab (Mathworks Inc, USA).

2.3 Wireless recording of place cells in vertical maze

The recording electrodes consisted of a bundle of eight Formvar insulated, 25 μm diameter nichrome wires (AM Systems, USA) assembled in a “Scribe” miniature on-head microdrive [3], as shown in Fig. 1. For implantation of the electrodes, six male Wistar rats (250–350 g) were anesthetized with sodium pentobarbital (50 mg/kg, i.p. Sigma. USA) and placed in a stereotaxic apparatus (David Kopf). The electrodes were inserted in area CA1 of the hippocampus (3.8 mm caudal to the bregma, 2.5 mm lateral to the midline, and ~1.8 mm ventral to the skull surface) [3] using a stereotaxic technique. The foot of the microdrive electrode assembly and seven anchoring screws were encased in dental acrylic. One screw served as electrical ground. All protocols were approved by the Medical Ethics Committee of the National Cheng Kung University Medical Collage and were performed according to the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health as well as the guidelines of the Animal Welfare Act.

To examine performance of the wireless system in the kind of situation that would be difficult with cable attachments we developed a vertical maze, as depicted in Fig. 2. One side wall was one-way transparent Plexiglas allowing video recording of animal’s behavior. The tunnel environment contained no barriers, permitting the rat to roam freely from one alley to another and turn around in either direction.
https://static-content.springer.com/image/art%3A10.1007%2Fs11517-008-0355-6/MediaObjects/11517_2008_355_Fig2_HTML.gif
Fig. 2

The isosceles trapezoid tunnel maze. The lower and upper tunnels are connected with stairs which are all made of Plexiglas. The rat enters the tunnel from one of two waiting chambers via a liftable ramp. Asterisks mark the location of visual cues at the positions at which food rewards were provided. Light emitting diodes (LEDs) located at the corners of the maze provided coordinates for localizing the rat’s position from LEDs mounted on the rat’s headstage

The rats had electrodes implanted and validated for recording of neuronal activities of place cells during open field foraging [3]. The recording electrode assembly was advanced slowly over the course of 4–8 days, (25–50 μm/day) until spikes of hippocampal place cell layer with amplitude greater than 100–150 μV and peak-to-peak duration about 300–500 μs were identified [5, 13]. The recorded signal was bandpass filtered with cutoff frequencies of 300 and 3,000 Hz for single-unit spike detection. The candidate spike was detected by threshold crossings of local energy measurement (root mean square) of which are larger than five times the standard deviation [4, 5]. To identify the single-unit activity, a sample of 3 ms (30 samples at a 10 kHz sampling rate) was chosen as a spike template which was calculated with a shuffled cross-correlation exceeding a selected threshold from the same unit.

Rats with good place cells recorded in the open field were habituated to the 3D maze by loading with the backpack with Bluetooth telemetry system and allowing them to run throughout apparatus. Scattered chocolate food reward was placed at the two specific positions marked with landmark cues to encourage exploration of the maze. The well-trained rats moved freely with the backpack without affecting behavioral performance in the 3D maze. After animal adaptation and system checking periods, a 30-min navigation experiment was performed for the recording of place cell responses.

The rat’s head location was monitored by a laterally mounted camera. The spatial position of rat’s head was referred to four LEDs located at four corners of vertical maze. The entire trapezoidal tunnel was divided into 2.5 cm × 2.5 cm bins perpendicular to the track for representing firing rate frequency as a function of position, i.e. the place field. Recordings of place cell activity were obtained from six separate sessions of 5 min free exploration each. The set of analyzed cells included only those cells that exhibited significantly place field characteristics in at least one the three sessions but not those were silent or fired weakly in the trapezoid tunnel.

3 Results

3.1 Structure of the Bluetooth-based wireless data acquisition system

The 8-channel AFE, digital control circuits and power supply circuits were laid on four-layer printed circuit board (PCB). The entire module was protected from power interference and other electromagnetic noise by a grounded copper foil wrapped around the PCB. The Bluetooth-based telemetry module was packed inside a 4 cm × 3.5 cm × 1 cm plastic box which weighed 25.3 g including all surface mount device components and rechargeable batteries. The entire wireless telemetry module could be easily carried on the back of rats, connected to the sensing wire electrodes and AFE short cable of less than 3 cm.

Counting the gain of amplification circuit of 2,500 and full scale output of the AFE at 3.7 V limited by the rechargeable battery, the maximal peak to peak voltage input to AFE is ±740 μV. Our selection of 12-bit A/D is adequate for single-unit recording without over-quantification of data using high resolution A/D, e.g. 16-bit or even higher, that would reduce the wireless data throughput. Generally, the amplitude of single unit waveform should exceed 3 times of the background noise for reliable spike discrimination. Thus, the minimal amplitude of detectable input single unit activity is suggested to be ≈50 μV.

Our Bluetooth module could bi-directionally transmit data at a maximal throughput rate of 120 kbps or equivalently 12-bit data sampling at 10 kHz, to a PC host over a range of 5 m. The effective transmission distance of 5 m was first validated by transmitting continuously a known data sequence, a numerical string from 0 to 4,095 generated by the MCU. The received data string was immediately reconstructed and observed on the screen of PC for verification. Although 10 m is supposed to be the maximum transmission distance of Bluetooth, our tests indicated that there was no data loss for transmission distance less than 5 m but transmission error rate increased dramatically to 7–10% at distances exceeding 8 m. The wireless module consumed 185 mW powered by rechargeable batteries which could continuously operate for up to 2 h of experiment. The detailed specifications of our wireless neural recording module are summarized in [3].

3.2 Single unit recording of rat navigating in vertical maze

Figure 3a shows a short section of typical place cells recording, depicting a series of 4 neuronal spikes within 20 ms (a “burst”) in the vertical maze. The amplitude of later spikes is smaller than that of earlier spikes, and the intraburst firing rate is less than 40 Hz, typical of “complex spike” bursts of hippocampal place cells [5, 11]. All of the spike waveforms from this cell and the average waveform are superimposed in Fig. 3b. The averaged peak-to-peak amplitude was 643.57 ± 174.86 μV with a CV of 27.17% and peak to trough width of greater than 300 μs.
https://static-content.springer.com/image/art%3A10.1007%2Fs11517-008-0355-6/MediaObjects/11517_2008_355_Fig3_HTML.gif
Fig. 3

Example of neuronal activity signal (extracellular action potential spikes) recorded from a rat navigating in the vertical trapezoidal tunnel environment. a A series of spikes at short interspike interval (a “burst”). b Overlay of all of the discriminated spikes, and averaged waveform (heavy line)

Example place-field analyses for recorded neuronal activity are shown in Fig. 4. Figure 4a, b shows the path traversed by the animal around the maze for one recorded cell, and the positions in which the cell fired spikes. The cell recorded during the session showed a strong place field associated with the rewarded position, firing exclusively at these specific places. Another cell also showing a place field associated with the rewarded location, but in a slightly different position is shown in Fig. 4c, d. In this recording session the rat sometimes turned around on the lower track at the place marked with triangles, and returned to the previously visited position. Notably, the place field characteristics of the cell were the same, independent of the direction the rat was heading when it entered the place field.
https://static-content.springer.com/image/art%3A10.1007%2Fs11517-008-0355-6/MediaObjects/11517_2008_355_Fig4_HTML.gif
Fig. 4

Examples of place fields for two cells recorded in the 3D trapezoidal tunnel. In a and c the path traversed by the rat during each recording session is shown in grey lines, and the places where the cell fired are shown as black squares. Stars indicate locations where chocolate was scattered as food award, and each triangle in c indicates the position at which the rat reversed direction of locomotion. The corresponding place field maps are shown in b and d, in which the instantaneous firing rate of the cell is averaged across all trials as a function of location. Grey-scale shows average firing rate value for each location in Hz

4 Discussion

In this study, we developed a flexible, miniaturized Bluetooth-based telemetry data acquisition system to record neuronal single-unit activity in free-moving rats. We confirmed its suitability for recording in environments in which a cabled system would be impracticable, by recording hippocampal place cells during navigation in a vertical maze in which the rat moved through a series of roofed spaces. The entire system was packaged in a backpack which is only about 10% of the body weight of an adult rat (250~350 g). The system was comfortably carried by the rats after appropriate habituation and caused no obvious inconvenience during their movements. The system also offers a long battery life meaning that prolonged, unsupervised recordings are feasible.

A major advantage of Bluetooth technology is the two-way transmission that allows for remote control of processes on the wireless unit. Our design provides a work-around for this problem by allowing 8 data channels of data to be previewed at 1 kHz sampling rate. This is sufficient to determine if there are spikes on any channel, and which has the best signal to noise ratio. Using the remote control function of the Bluetooth module, commands can be sent to the recording unit to set the sampling rate of the chosen channel to 10 kHz. The Bluetooth telemetry system produced equally strong signals regardless of the animal’s position inside a 5-m-diameter field. Since the effective distance of data transmission depends on the experiment environment, a validation of data transmission should be performed in each experimental setting. To ensure there is no data loss during experiment, the four most significant bits (MSB) of transmitted 16-bit data also served as time marker. Once there is data loss, the data should be resent. If too many transmission errors occurred and caused over run in data acquisition process, the entire experiment should be restarted and rechecked.

During the development of this Bluetooth-based telemetry system, we observed that the recorded signals are highly susceptible to background noise, including 60 Hz power line interference and Bluetooth hopping frequency. These interferences significantly affected the signal to noise ratio of the recorded signal. The major reasons are the small amplitude of extracellularly recorded potentials from single neuron and the high impedance of wire electrode, usually around 300 to 500 kΩ, which make it susceptible to noise. These interferences were not evident in previous pilot study of ECG recoding [14] and were not clearly mentioned in other studies using wireless telemetry in which the signal was usually larger and the electrode impedance was smaller compared to wire electrode recording of single-unit potentials.

Several measures including proper circuit layout design and grounding arrangement were essential to reduce these interferences and to improve the SNR of recorded signals [3]. To avoid electromagnetic interference, the AFE circuit should be placed close to the microdrive electrodes and far away from wireless transceiver during PCB layout. Appropriate grounding for the analog ground and digital ground of data acquisition system should be separated to avoid the coupling of high frequency background noise from the digital part to the analogue one. Our results indicated that a floating battery-powered telemetry system with separated power sources for analogue and digital module as well as Bluetooth transceiver can alleviate the coupling of RF interference along the ground loop. Shielding to the entire wireless module and single-end grounding to the transmission line are critical to minimize the 60 Hz interference. Having taken all the above precautions, signal to noise ratio of 3:1 was able to be achieved while rats moved freely in the tunnel.

By integrating the neuronal recording and the spatial information derived from video recordings, the characteristics of hippocampal place cells can be investigated for animal navigating in a space. It is known that hippocampal place cells typically change their firing properties unpredictably between two distinct environments [9]. However, most studies on place cells have been restricted to essentially 2D environments, leaving unanswered the question of whether place cells are insensitive to changes in the z axis, the vertical orientation of space. However, a major limitation of current studies on place field in 3D space by using wired recording is the inability to record neural activity in environments in which animals need to go through enclosed spaces, tunnel wall, or branches with obstructions [8]. In this study, we demonstrated the potential utility of our wireless system by continuously recording hippocampal neural activity in an enclosed, vertical maze. This system thus enables a vastly richer array of behavioral test scenarios than those are possible with cabled devices. In the future, upgrading of the Bluetooth protocol to version 2.0 and peripheral communication interface from UART to USB 2.0 may allow wireless transmission data rate up to 3 Mbps which would provide simultaneous recording of 8-channel single unit potentials at a sampling rate up to 30 kHz and allow multiple tetrode recording, which is particularly useful to improve single unit discrimination in densely packed cell layers of the hippocampus.

Acknowledgments

This project was partially supported by grants from the National Science Council, Taiwan, R.O·C. (NSC93-2213-E-006-046) and from the National Health Research Institute of R.O·C. under contract no. NHRI-EX 95-9524E1.

Copyright information

© International Federation for Medical and Biological Engineering 2008