Open-source Python software for analysis of 3D kinematics from quadrupedal animals. (May 2019)

• Record Type: Journal Article
• Authors: Maghsoudi, Omid Haji; Vahedipour, Annie; Hallowell, Thomas; Spence, Andrew
• Is Part Of: Biomedical signal processing and control. Volume 51(2019)
• Page Start: 364
• Page End: 373
• Publication Date: 2019-05
• Subjects: Neuroscience -- Biology -- Biomechanics -- Python -- Kinematics -- 3D Reconstruction; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2019.02.024
• Languages: English
• ISSNs: 1746-8094

Abstract

Abstract: Locomotion is key to survival, yet it can be disturbed by injuries, diseases, and aging. Therefore, it is important for researchers in biology, neuroscience, biomechanics, and further afield. Locomotion is frequently quantified using kinematic data, and quadrupeds including rodents are medical model animals used worldwide. Kinematics typically result from the tracking of some landmarks, often joint centers. Although commercially available systems and software exist for the analysis of these kinematic data, they are expensive, often restricted to use with a given apparatus, and may be inextensible. Therefore, there is a need for an open source tool to analyze kinematic data. We present a Python software to address this need. It uses 2D coordinates from four cameras and DLT coefficients from the calibrated volume to generates 3D coordinates[1] . A method is presented to modify the knee and elbow joint positions in 3D. Then, kinematic features are extracted, and they are sorted in a time series format to plot a summary of a study. In addition, we generate videos from the tracked points, 3D reconstruction of the points, showing joint angles for eight joints, the location of animal on the belt, and the animal's speed on the belt. The software has been evaluated by eight trials to show the importance of the work. The 3D reconstruction error, having an average of 7.36 pixels, was calculated for the markers. The presented program can be used in different fields. It will encourage the researchers to design the studies based on their needs because they can change the setup in any required conditions while they can extract the kinematic data.