Automated Gait Analysis

Abstract

Gait analysis is the systematic study of walking patterns. Typically, expensive and cumbersome marker-based methods, in conjunction with multiple infrared cameras, are used to produce kinematic data that concisely quantify the walking behaviour of a person. From the kinematic data, specialists can better understand an individual’s walking behaviour and diagnose gait-related conditions. This research aims to develop an automated alternative to marker-based methods for gait analysis. An automated method based on artificial intelligence is proposed that can produce kinematic data consisting of varying left and right joint angles for hips and knees. It obtains all of this from videos capturing the side and front views of a walking subject. The method makes use of pose estimation, as well as a pipeline of techniques for calculating and processing kinematics. Included in the pipeline is a procedure based on ankle displacement for sampling walks into cyclic periods known as gait cycles. Video data for the automated method, as well as marker data for a well-established marker-based method, were collected at a biomechanics lab, simultaneously for comparison. Relative to the marker-based method, the automated method achieved a minimal error of 3.92 degrees for left and right joint angles of hips and knees. Binary classification was investigated with the kinematics of the automated method, in samples of left and right gait cycles. Support Vector Machines achieved a 79% accuracy in detecting whether the walk of a person is normal or abnormal. The automated method was proven to be more advantageous than marker-based methods for data collection and processing. It reduced the effort and financial investment needed, potentially leading to a broader diffusion in the health care and research community. In the future, through the proposed automated method, a classifier can be trained with sufficient kinematic data to accurately deduce gait-related conditions and minimise the involvement of a specialist for diagnosis.