ABSTRACT
In rodents, detection and quantification of motor impairments is difficult. The traction test (inverted grid with mice clinging to the underside) currently has no objective rating system. We here developed and validated the semi-automatic MATLAB script TracMouse for unbiased detection of video-recorded movement patterns. High precision videos were analyzed by: (i) principal identification of anatomical paw details frame-by-frame by an experimentally blinded rater; (ii) automatic retrieval of proxies by TracMouse for individual paws. The basic states of Hold and Step were discriminated as duration and frequency, and these principle parameters were converted into static and dynamic endpoints and their discriminating power assessed in a dopaminergic lesion model. Relative to hind paws, forepaws performed ~4 times more steps, they were ~20% longer, and Hold duration was ~5 times shorter in normal C57Bl/6 mice. Thus, forepaw steps were classified as exploratory, hind paw movement as locomotive. Multiple novel features pertaining to paw sequence, step lengths and exploratory touches were accessible through TracMouse and revealed subtle Parkinsonian phenotypes. Novel proxies using TracMouse revealed previously unidentified features of movement and may aid the understanding of (i) brain circuits related to motor planning and execution, and (ii) phenotype detection in experimental models of movement disorders.
