Pipeline Validation for Rodent Gait Analysis Using deeplabcut

This study aimed to integrate deeplabcut (dlc) with our Automatic Gait Analysis Through Hues and Areas (AGATHA) algorithm. Prior work with AGATHA shows that it can be used to understand spatiotemporal gait adaptations in multiple disease models. However, AGATHA cannot detect kinematic variables, like joint angles, which dlc was designed to measure. Here, these two approaches are integrated, and the gait variables that can be achieved with both methods are compared. To train dlc, hand digitization of high-speed videos was conducted to estimate the location of several key anatomical markers; then a neural network within dlc was used to automate the digitization of these same points in subsequent videos. A matlab pipeline was developed to calculate average stride profiles for toe height, back angle, and midfoot angle from dlc coordinates. Then, 418 videos of naïve Sprague-Dawley rats (12 w.o., n = 18) walking unprompted across an arena were collected. These videos were analyzed using dlc and AGATHA. For velocity, hind limb duty factor, stride length, and step width, variability was larger in dlc than in AGATHA. However, when used in conjunction, dlc and AGATHA have strong complementary datasets, where AGATHA can provide spatiotemporal and dynamic measures, and dlc can provide kinematic measures that AGATHA cannot currently measure. Decisions on whether to use AGATHA alone, dlc alone, or AGATHA-dlc in tandem are thus dependent on the gait variables being evaluated.