Simultaneous, Non-Contact and Motion-Based Monitoring of Respiratory Rate in Sheep Under Experimental Condition Using Visible and Near-Infrared Videos

Abstract

The validation of methods for understanding the effects of many diseases and treatments requires the use of animal models in translational research. In this context, sheep have been employed extensively in scientific studies. However, the imposition of experimental conditions upon these animals may result in the experience of discomfort, pain, and stress. The ethical debates surrounding the use of animals in research have resulted in the adoption of Directive 2010/63/EU. The present study proposes a non-contact method for monitoring the respiration rate of sheep based on video processing. The Detecron2 model was trained to segment the sheep's body, abdominal, and facial regions in the video frames. A motion-tracking algorithm was developed to assess abdominal movement associated with the sheep's respiratory cycle. The method was applied to videos of Rhön sheep under experimental and housing conditions, utilising two types of cameras to assess the effectiveness of the proposed approach. The mean average error (MAE) obtained was 0.79 breaths/minute for the visible and 1.83 breaths/minute for the near-infrared (NIR) method. This study demonstrates the feasibility of video technology for simultaneous and non-invasive respiration monitoring, being a crucial parameter for assessing the health deterioration of multiple laboratory animals.

Keywords: animal welfare; refinement; respiration rate; sheep model; video signal processing; welfare assessment.

Figure 1

Schematic overview of the methods employed to extract RR in sheep using RGB and NIR videos.

Figure 2

Sample of a frame collected during the experiments with the RGB camera. The figure shows a Rhön sheep resting with a cast on its leg and scrubs. A drawing also shows the regions of the sheep’s thorax corresponding to the R1, R2, R3, and total regions, which were later used to evaluate RR extraction.

Figure 3

Schematic view of the second study stage showing the areas where the NIR videos were taken. The top of the figure shows which areas correspond to Area1 and Area2. Then, at the bottom, the time ranges at which sheep were at rest (dark blue). It is also presented the separation between day (7 a.m.) and night (7 p.m.).

Figure 4

Images used during the training process. (a) Sheep generated artificially using the method developed by DALL-E. (b) Visualisation of the LabelMe tool with an annotated sheep image, where the colour representations correspond to the red area denoting the sheep, the green area representing the chest, and the yellow area indicating the face.

Figure 5

(a) Signal obtained through the processing of methods for acquiring the respiratory cycle in an RGB video. The graph shows the periods of inspiration and expiration caused by the thoracic movement tracked by the FP across the frames, where peaks correspond to inhalation and valleys to the exhalation of the analysed sheep. There is also a representation of the sliding window to illustrate the calculation interval of RR. (b) RR obtained through the processing of the respiratory cycle signal. The graph displays the rate curve and an average line over time in seconds.

Figure 6

IoU in the validation set during the iterations of the Detectron2 training process.

Figure 7

Process of delimiting the sheep’s thoracic region to acquire the RR using the generated Detectron2 model. (a) RGB original image. (b) RGB binary mask. (c) RGB segmented thoracic region. (d) NIR original image. (e) NIR binary mask. (f) NIR segmented thoracic region.

Figure 8

(a) Correlation plot comparing RR_Ref and RR_Video distinguishing the three ROIs. The ID specified in the plot refers to the processed video. (b) Box plot showing the relative errors obtained by each ROI and the total area.

Figure 9

(a) Correlation plot comparing RR_Ref and RR_Video acquired by processing NIR videos. (b) Bland-Altman plot comparing RR_Ref and RR_Video acquired by processing NIR videos.