Scratching-proxy occupancy and open-field locomotion in DCP-group mice: an exploratory three-dimensional pose analysis

We asked whether excluding a pose-based scratching proxy reduced the open-field activity difference between DCP-group mice and controls. Ten mice treated with diphenylcyclopropenone (DCP) and ten independent controls were recorded for 600 s with sixteen-keypoint three-dimensional pose tracking at 30 frames per second. A hind-paw-to-head proximity rule labelled 0.63% of control frames and 18.71% of DCP-group frames (Welch p = 4.93 × 10⁻⁴, Hedges g = +2.26). Total distance was 34.7% lower in the DCP group; after excluding labelled frames and normalising retained time, the difference was 31.9%. Thigmotaxis was 87.10% versus 85.21% (q = 0.721, g = −0.21; 95% CI [−1.11, +0.68]). DCP-group mice also had shorter body-axis estimates, less detected rearing and poorer pose consistency. The detector remains a proxy pending comparison with annotated video. Recording batch coincides with group, and the DCP schedule is incomplete. Excluding detected frames therefore leaves a descriptive locomotion difference, without identifying a treatment effect or the fraction mediated by scratching.