DeepLabCut increases markerless tracking efficiency in X-ray video analysis of rodent locomotion

Area of Science:

• Biomechanical analysis
• Animal models in research
• Machine learning applications

Background:

• Quantifying skeletal movements in rat models is crucial for studying human disease and injury.
• Current methods like optical video analysis and manual X-ray analysis have significant limitations, including skin movement inaccuracies, invasiveness, and time consumption.

Purpose of the Study:

• To evaluate the efficacy of DeepLabCut, a machine learning tool, for automated, markerless tracking of skeletal kinematics in bi-planar X-ray videos of locomoting rats.
• To assess the accuracy, precision, and time efficiency of DeepLabCut compared to traditional methods.

Main Methods:

• Trained DeepLabCut models on 590 pairs of video frames to identify 19 unique skeletal landmarks of the rat pelvic limb.
• Assessed accuracy and precision by comparing machine-identified landmarks to manually labeled counterparts.
• Quantified time savings achieved through automated analysis.

Main Results:

• Machine-identified landmarks showed a deviation of 2.4±0.2 mm from manual labels across 1710 landmarks.
• DeepLabCut reduced analysis time by 1627-fold compared to manual labeling.
• Demonstrated high accuracy and precision in markerless tracking of skeletal movements.

Conclusions:

• DeepLabCut provides an accurate, precise, and highly efficient method for analyzing rat locomotion kinematics from X-ray videos.
• This markerless approach overcomes the limitations of existing methods, enabling large-scale data processing.
• Facilitates faster and more accessible research in animal models without invasive procedures.