Deep-Learning Based Multi-Joint Synchronous Tracking for Objective Quantification of Hindlimb Locomotor Kinematics in Rats

We developed an AI-driven treadmill to track rat movements, helping us precisely evaluate spinal cord injury recovery. Unlike traditional footprint methods that means complex joint economics, our mock AI system directly tracks multi-joint movements. To begin, turn on the industrial personal computer and launch the video acquisition and analysis software.

Position the camera lens perpendicular to the treadmill longitudinal axis to obtain a strictly lateral view, maintaining a horizontal working distance of 15 centimeters from the treadmill belt plane. Next, record the body weight of each rat with an electronic scale. Include only rats with matched body weights to minimize body size effects.

Put on gloves and gently grasp the rat. Loop the elastic chest strap around the front armpit and attach it to the adjustable slide rail. Then adjust the strap so that one finger can be inserted without restricting breathing.

On the touch control screen, set the speed to 150 millimeters per second and the incline to zero degrees. Conduct a 10-minute acclimation session to adapt the rat to the body weight support level while monitoring the rat for signs of stress and exhaustion, such as reluctance to move and prolonged defecation. Confirm successful acclimation when the rat maintains a continuous and uniform stride for at least 60 seconds without paw dragging and with a naturally hanging tail.

Exclude animals that fail to meet these criteria after the maximum acclimation period. Enter the parameters for the formal experiment on the touchscreen. Sequentially input the desired speed and incline, then select the treadmill belt direction.

After stabilization, select Start Recording to begin data acquisition. Continuously capture at least five complete gait cycles. At the end of each trial, reduce the speed to zero millimeters per second.

Unclip the chest strap and return the rat to its corresponding cage. Extract video segments of approximately 10 seconds that contain the target gait with at least 10 stable gait cycles. Register the rat information, including identification number, group, and experimental conditions.

Import the MOV file into the analysis software. Normalize each gait cycle from zero to 100%to standardize cycle length. Generate representative images to illustrate gait dynamics.

Export the spatial position coordinates of each joint over time as a csv file for further analysis. After completing animal modeling, implant the electrophysiological recording device. For brain signal recording, place electrodes on the skull surface, epidural space, or cerebral cortex to record brain signals.

For spinal cord recording, insert recording electrodes into the epidural space of the intervertebral foramen. For electromyography recording, bury bipolar silver wires into the target muscles to record muscle electrical activity. Allow the animal to recover for five to seven days after implantation.

Check the wound and gait daily to ensure no signs of infection or pain before treadmill testing. Prepare the equipment and animal before the synchronized experiment. Then synchronize the electrophysiological data acquisition with the movement video to ensure both share the same timestamp.

Align the neural signals with corresponding video frames during analysis to visualize electrophysiological patterns at different gait phases. The rats with spinal cord injury exhibited significant loss of swing and irregular iliac displacement curves. Spinal cord injury rats showed increased joint angle fluctuations compared with healthy rats.

The joint range of motion heat map showed a widespread decrease in color scale in spinal cord injury rats compared with healthy rats, while the trajectory activity map revealed a significant reduction in the range of motion of each joint, along with impaired movement continuity. For spinal cord injury rats, the point cloud map showed increased dispersion and a leftward shift of the center of mass. The height waterfall chart shows a bimodal distribution along the x-axis between groups with a general difference drop in the middle of the movement.

For the y-axis, a difference drop is observed at the beginning of the movement. The overall difference distribution is chaotic, indicating abnormal movement phases in the spinal cord injury rats compared to healthy rats. The vertical height peak distribution of the toes was shifted to the right in spinal cord injury rats.

The velocity range chart showed that the combined velocity range in each frame was narrowed in the injury group. Furthermore, the spinal cord injury rats also exhibited a reduction in the phase plane area, the estimated peak propulsion force index, and movement smoothness indicators. Our system allow researchers to precisely measure the multi-joint trajectories, the force distribution, and the movement smoothness in real time.

An important consideration is thoroughly acclimating the rat to the treadmill system to ensure stable and analyze gait patterns. We can synchronize the system with EMG or EEG to decode the signals of mechanisms behind this movement.