Wireless inertial measurement of head kinematics in freely-moving rats

Abstract

While miniature inertial sensors offer a promising means for precisely detecting, quantifying and classifying animal behaviors, versatile inertial sensing devices adapted for small, freely-moving laboratory animals are still lacking. We developed a standalone and cost-effective platform for performing high-rate wireless inertial measurements of head movements in rats. Our system is designed to enable real-time bidirectional communication between the headborne inertial sensing device and third party systems, which can be used for precise data timestamping and low-latency motion-triggered applications. We illustrate the usefulness of our system in diverse experimental situations. We show that our system can be used for precisely quantifying motor responses evoked by external stimuli, for characterizing head kinematics during normal behavior and for monitoring head posture under normal and pathological conditions obtained using unilateral vestibular lesions. We also introduce and validate a novel method for automatically quantifying behavioral freezing during Pavlovian fear conditioning experiments, which offers superior performance in terms of precision, temporal resolution and efficiency. Thus, this system precisely acquires movement information in freely-moving animals, and can enable objective and quantitative behavioral scoring methods in a wide variety of experimental situations.

Figure 1. Overview of the wireless inertial measurement system.

(A) Inertial measurement unit (IMU) and its main components. (B) Left: Photograph of an adult rat wearing the IMU. Right: sketch showing the directions of linear accelerations (a_x, a_y, a_z) and angular velocities (ω_x, ω_y, ω_z) measured by the sensor, with respect to the animal’s head. (C) Simplified diagram of the circuit managing the acquisition of inertial data and inbound/outbound synchronization on the IMU. (D) Schematic of the whole system, that comprises an IMU, a PC equiped with a Bluetooth USB dongle, a data acquisition (DAQ) board, a custom IR LED controller for inbound synchronization and a custom low-latency ISM receiver for outbound synchronization. The transmission range is indicated for each wireless communication channel. BT: Bluetooth; IR: infrared; ISM: industrial, scientific, medical; MCU: microcontroller unit.

Figure 2. Bluetooth transmission of inertial data.

(A) Left: schematic of the recording configuration. Right, top: accelerometer data along the 3 axes (a_x, a_y and a_z) were plotted against their reception time by the client software. Note the presence of occasional large gaps between successive data frames. Right, bottom: corresponding instantaneous data reception rate (bin = 10 ms). (B) Density histogram of time intervals between successive data frames in a 1 h recording. Inset: magnified view of the area delimited by a red dotted line. Note the presence of long intervals (>10 ms). (C) Left: schematic of the experiment used to measure transmission delays. A 1.5 V battery was hit against a metal plate, creating a voltage difference across a resistor (recorded using a DAQ) and a mechanical vibration (recorded using the wireless IMU or a wired IMU). Right: For each shock (here one shock is shown as an example), the first points significantly deviating from baseline were identified in the electrical (V, top) and inertial (a_z, bottom) signals. The interval between the two points (red points in blown up traces) was taken as a measure of the data transmission delay. (D) Normalized histograms of transmission delays for the wired IMU (green) and the wireless IMU (blue), calculated using a total of 250 shocks (bin = 2 ms).

Figure 3. Bidirectional synchronization using two separate wireless channels.

(A) Left: schematic of the acoustic startle experiment. Bursts of white noise (25 ms) were generated by a DAQ and played by a loudspeaker. At the onset of each stimulus, a synchronous 5 V signal was used to gate the emission of an IR synchronization signal. Right: startle responses recorded in one animal. Linear acceleration along the 3 axis (a_x, a_y and a_z) was aligned on the onset of the IR event and averaged. Superimposed gray lines represent individual sweeps and color lines represent average linear accelerations. (B) Left: schematic of the closed-loop, motion-triggered experiment. The IMU was configured to emit RF power when yaw angular speed towards the left exceeded 200°/s. The output of the RF receiver was monitored through the digital input channel of a DAQ and conditioned the execution of an analog output task. The analog output was a simple 5 V command that was used to gate the emission of IR signals. Right: Yaw angular velocity (ω_z) and outbound/inbound synchronization information recorded by the IMU during a 12 s period. The dashed green line represents the angular velocity threshold above which RF power was emitted by the IMU. The black and red lines below represent the states of the booleans reporting the detection of an angular velocity value exceeding threshold (Threshold crossing, black) and the presence of an IR signal (State of inbound synchronization, red).

Figure 4. Head kinematics during free ambulation in rats.

(A) Top: average density histogram of head angular speeds (, , ). Bottom: average density histogram of the head linear accelerations. The different mean values of a_x, a_y and a_z reflect the influence of gravity. (B) Average power spectral density histograms for accelerometer and gyroscope data. Average histograms in (A,B) were calculated from 19 rats.

Figure 5. Tracking head posture in freely moving rats after a unilateral vestibular lesion.

(A) Example traces showing the slow (<2 Hz) components of linear acceleration (alp_x, alp_y and alp_z, colored lines) superimposed on the raw linear acceleration (gray lines). These filtered signals capture the variations of linear acceleration due to reorientations of the head relative to gravity. (B) Left: alp (the vector defined by alp_x, alp_y and alp_z) is an approximation of gravity in the sensor’s coordinates. Right: density map of the different orientations of alp encountered during a 45 min recording session. (C) Example showing the effect of a unilateral vestibular lesion in one rat. Top: alp orientation density maps. Bottom: corresponding average head postures. (D) changes in the average pitch and roll angles (as defined in C) across sessions. Shaded areas represent the SEM. Symbols indicate statistically significant differences from angle values calculated before lesion (^§p < 0.01, ^#p < 0.001, paired t-test).

Figure 6. Automatic detection of behavioral freezing in a classical fear conditioning experiment.

(A) Schematic of the experiment (see also Fig. S9C). The conditioned stimulus (CS) was a 30 s white noise. Fear conditioning (day 2, CS+ electric shock) and testing (days 3–7, CS only) occurred in two different contexts (A,B). The CS was presented 6 times during each of the 5 testing sessions, together with a synchronous IR signal that was recorded by the IMU (IR inbound sync.). (B) Example traces showing head angular velocities (top, colored traces) and angular speed (bottom, black trace) during a trial, defined as a period encompassing one CS presentation and the 30 s before (Pre-CS) and after (Post-CS). The dashed line represents an immobility detection threshold set at 12°/s. (C) Fear extinction profile for one rat. The average per session immobility scores (±SD as shaded area) were calculated using the observer’s data (black line) and the “discrete” automatic scoring method (red line) for the intervals before (Pre-CS), during (CS) and after (Post-CS) CS presentation. The fraction of time spent immobile was calculated using the “continuous” automatic scoring method (blue line). The horizontal dashed line represents the fraction of time spent immobile during the first exposure to context A (day 1). No significant differences were found between curves for a given extinction session (p always greater than 0.14, unpaired t-test). For days 4 to 7 (second to fifth extinction session), asterisks (color corresponding to the type of scoring technique) indicate whether the freezing score was significantly different from day 3 (first extinction session), with *p < 0.05, **p < 0.01 and ***p < 0.001 (unpaired t-test).