Grimace analysis, simplified

PainFace is a software platform for measuring spontaneous pain from mouse facial expressions. The software runs in a browser window. PainFace works with videos of C57BL/6 mice in controlled, well-illuminated conditions. Future builds are in development for white- and agouti-coated mice.

McCoy et al. (2024) Pain. Please cite this reference when PainFace is used to facilitate your research.

A username and password is required to use PainFace.

Click "No account? Sign up" link on the PainFace login screen to set up an account.

Imaging Conditions

Acquire images or videos (must be mp4 format) of mice under well-illuminated conditions with minimal shadows and no reflections. See download section for a sample video. Users are strongly encouraged to replicate the imaging conditions shown in this sample video and in our McCoy et al paper. This includes imaging angle relative to the mouse, light levels, and 1080p (high definition) color video. These conditions were used to generate the ground-truth data, so score accuracy may be low if other imaging conditions are used.

Recommended behavioral chamber: We developed a mouse behavioral chamber that is optimized for automated grimace analyses with PainFace. See McCoy et al (2024) for details.

Recommended camera: Sony Handycam CX405. Use the following settings. Shooting mode=Movie. Camera/Mic.-Manual Setting: White balance=auto; Exposure=manual; Focus=auto; IRIS=auto. Camera/Mic.-Camera Settings: Scene selection=auto, Picture effect=off; SteadyShot=auto; DigitalZoom= off; Auto back light=on. Image quality/size: Frame rate=30p, Dual video REC=off, File format=XAVC S HD. Setup-Connection: TV Type=16:9, HDMI resolution=1080p, CTRL for HDMI=on.

Use a 150 Mb/s (or higher) transfer speed SDXC card to prevent frame loss when captured at 30 fps.

Data

Videos that users upload to PainFace may be used to improve the accuracy of the machine learning algorithm.

Acknowledgments

This work was supported by the National Institutes of Health / The National Institute of Neurological Disorders and Stroke (R01NS114259).

References

Langford DJ et al (2010) Nat. Methods. PMID: 20453868. Original description of the Mouse Grimace Scale (MGS) and human scoring of facial action units.

Tuttle AH et al (2018) Mol. Pain. PMID: 29546805. First use of machine learning to automate Mouse Grimace Scale (MGS) scoring.

McCoy et al (2024) Pain. PMID: 39024163. Original PainFace paper, describes development, validation, and use of the software.

FAQs

1. Can I use PainFace with mice in their home cage? We have not validated the accuracy of the PainFace MLA when mice are imaged under these conditions.

2. Can I use monochrome (black and white) images/videos? No.

3. My original video is clear and well-focused, but after importing the video into PainFace, some of the individual image frames are distorted / blurry / have lines. How do I fix this? Make sure you are capturing videos at 30 fps directly to mp4 format onto a 150 Mb/s (or higher) SDXC card. We found that distortions can occur if you acquire videos in a different format and then use conversion software to generate a mp4.

4. Does PainFace work with white, agouti, or grey coated mice? The software has not been trained to work with mice of any other coat color besides black, so accuracy may be lower.