Abstract
Contactless monitoring of laboratory rodents is essential for improving animal welfare and ensuring the integrity of physiological data. This paper presents a non-invasive monitoring system utilizing 60 GHz mm-wave Frequency-Modulated Continuous Wave (FMCW) radar. The system achieves 99% accuracy in respiration rate detection and 93% accuracy in animal localization. By eliminating the need for invasive implants and minimizing stressful human handling, this technology provides a high-fidelity solution for longitudinal biomedical research in home-cage environments.
Introduction
In biomedical research, monitoring the health and behavior of laboratory rodents often necessitates a compromise between data quality and animal welfare. Traditional methods, such as surgical implants or physical restraint, introduce significant stress—a major confounding variable that can spike heart rates and corticosterone levels, potentially masking experimental effects. While camera-based systems offer a non-invasive alternative, they are frequently limited by low-light conditions, bedding obstructions, and the visual similarity of group-housed animals.
As shown in the comparative analysis above, this work addresses these challenges by providing a multi-organism, non-invasive solution that supports both localization and respiratory monitoring in dynamic environments.
Methodology
The system utilizes an Infineon BGT60 series 60 GHz FMCW radar, which allows for the simultaneous determination of target range and velocity. To balance the requirements of tracking rapid locomotion versus detecting minute physiological tremors, the researchers implemented a "Dual-Mode" strategy:
- Movement Configuration: Optimized for tracking position and activity levels at a 40 Hz frame rate.
- Vital Sign Configuration: Optimized for micro-movements at a 400 Hz frame rate, achieving a 2 µm distance resolution to detect chest wall displacements.
Signal Processing Pipeline
To extract clean signals from "heavy clutter" environments like plastic cages, the team employed a Running Average Filter (RAF) to subtract static reflections. Respiration rates (RR) were derived from phase extraction at the specific range bin of the animal. For heart rate (HR) detection, the system analyzes the second derivative of displacement (acceleration) to isolate the sharp, impulsive mechanical signals of the heart from the larger, slower movements of the lungs.
Experimental Validation
The system was validated through two primary stages:
- Controlled Validation: An "Agar puppet" simulating animal tissue was used to test radar precision against known vibration frequencies.
- In-Vivo Evaluation: Two female Swiss Webster mice were monitored in a specialized interlinking cage setup. High-resolution video recordings served as the ground truth for both localization and respiration rates.
Key Results and Discussion
- Respiration Accuracy: The radar system achieved a 99% accuracy rate for RR compared to ground truth, with a mean error margin of ±2 bpm.
- Localization and Sensitivity: The system correctly identified the mouse's zone (Near, Middle, or Far) with 93% accuracy and detected motions as small as 0.1 mm/s.
- Hardware Efficiency: Operating at less than 5 mW in duty-cycled mode, the hardware is suitable for long-term, battery-powered deployment.
- Radome Optimization: Analysis determined an ideal distance of 3.5 cm between the radar and the cage wall to minimize signal reflection and maximize energy transfer to the subject.
Conclusion and Future Work
This 60 GHz radar system represents a significant advancement in unobtrusive animal monitoring. While the system excels at respiration and activity tracking, heart rate monitoring in freely moving animals remains a challenge due to signal-to-noise ratios during active behaviors like sniffing. Future research will focus on scaling the system for larger group-housed cohorts and utilizing machine learning to classify complex behaviors from micro-Doppler signatures.
Summary updated