Fig 1.
Tracking the facial features of a head fixed mouse.
A) Illustration of the different facial features that are tracked and their change and positional change from relaxed to stressed. B) Screen capture of Video footage from a camera facing head fixed mouse with the tracking points overlaid. C) Positional frame summary of the movement range for each facial feature.
Fig 2.
Schematic of the live video transfer from the experimental computer to the analysis computer and the return of the stress evaluation results.
A) Depiction of the 3 different stages of stress shown via video capture and the corresponding icon. B) Schematic representation of the camera setup and the data transfer between the experimental computer and analysis computer.
Fig 3.
Evaluation of the effect of facial feature weight adjustment.
A) Three Logarithmic graphs of the accuracy for each facial feature. Dotted line, accuracy threshold. The accuracy of the Ear Left, Eye Right, Eye Left and Nose are close to each other and overlap in the graph. The top graph represents a mouse with all facial features visible. The middle shows the effect of temporal obstruction, moving objects between the mouse and the camera. The bottom shows the effect of permanent obstruction of a facial feature. B) Comparing the standard weight distribution and the adjusted weight distribution. C) Calculated stress level of the same mouse using equal weight distribution and adjusted weight distribution. D) Effect of removing facial features on stress level evaluation. The Y-axis shows the deviation in stress level compared to the baseline using all five features.
Fig 4.
Comparing the stress evaluation results using different methods.
A) Estimated stress level based on the facial feature evaluation of the participants. B) Measurement of the corticosterone metabolites for each of the mice shown in the 7 videos. C) Comparing the previous 2 methods with the results from MouseCare. Inter-rater reliability among the senior researchers was assessed using Cohen’s kappa coefficient to ensure consistency, and the Corticosterone metabolites were normalized. D)Calculation of the Pearson correlation coefficients between the three methods for each mouse. The stool samples were normalized. The X and Y axis represent the stress level on a scale from 1-10. In gray is the confidence interval. The Pearson correlation coefficient is labeled “r” and the p value is labeled as “p”. E) Using the Monte Carlo Simulation of the 7 mice and the three different methods of stress evaluation using 10000 simulations.
Fig 5.
Impact of habituation on the response to external stress stimuli.
A) Evaluation of the facial features between naive and trained mice in response to external stress stimuli using MouseCare. B) One time external stress stimulus after different days of habituation. C) Bar chart of the facial area of single mice at the day of stress introduction.