Types of anomalies in two-dimensional video-based gait analysis in uncontrolled environments

Abstract

Two-dimensional video-based pose estimation is a technique that can be used to estimate human skeletal coordinates from video data alone. It is also being applied to gait analysis and in particularly, due to its simplicity of measurement, it has the potential to be applied to gait analysis of large populations. However, it is considered difficult to completely homogenize the environment and settings during the measurement of large populations. Therefore, it is necessary to appropriately deal with technical errors that are not related to the biological factors of interest. In this study, by analyzing a large cohort database, we have identified four major types of anomalies that occur during gait analysis using OpenPose in uncontrolled environments: anatomical, biomechanical, and physical anomalies and errors due to estimation. We have also developed a workflow for identifying and correcting these anomalies and confirmed that this workflow is reproducible through simulation experiments. Our results will help obtain a comprehensive understanding of the anomalies to be addressed during pre-processing for 2D video-based gait analysis of large populations. Author summary Gait is one of the important biomarkers of numerous health conditions. With developing mobile health technologies, it is becoming easier to measure our health. However, establishing evidence is a critical issue to providing preventive medicine, we need to be able to collect data from a large population. Two-dimensional video-based pose estimation can be a solution for the gait analysis of such a population. However, the technical accuracy and limitations of this analysis method have not yet been sufficiently discussed. In this study, by analyzing the largest database currently available, we systematically identified four types of technical anomalies that occur during gait measurement: anatomical, biomechanical, and physical anomalies and errors dues to estimation. We have also shown how to deal with these issues and made solutions available as software so that researchers can reproduce them. In the future, increasing numbers of studies will use 2D video-based pose estimation to research health-related gait among large populations. We believe that our work will provide a guideline for researchers and clinicians involved in these studies to discuss design and algorithms.