Deep Learning Techniques for Improving Digital Gait Segmentation

Abstract

Wearable technology for the automatic detection of gait events has recently gained growing interest, enabling advanced analyses that were previously limited to specialist centres and equipment (e.g., instrumented walkway). In this study, we present a novel method based on dilated convolutions for an accurate detection of gait events (initial and final foot contacts) from wearable inertial sensors. A rich dataset has been used to validate the method, featuring 71 people with Parkinson's disease (PD) and 67 healthy control subjects. Multiple sensors have been considered, one located on the fifth lumbar vertebrae and two on the ankles. The aims of this study were: (i) to apply deep learning (DL) techniques on wearable sensor data for gait segmentation and quantification in older adults and in people with PD; (ii) to validate the proposed technique for measuring gait against traditional gold standard laboratory reference and a widely used algorithm based on wavelet transforms (WT); (iii) to assess the performance of DL methods in assessing high-level gait characteristics, with focus on stride, stance and swing related features. The results showed a high reliability of the proposed approach, which achieves temporal errors considerably smaller than WT, in particular for the detection of final contacts, with an inter-quartile range below 70 ms in the worst case. This study showes encouraging results, and paves the road for further research, addressing the effectiveness and the generalization of data-driven learning systems for accurate event detection in challenging conditions.

Non-Linear Template-Based Approach for the Study of Locomotion

“…Reported scores for ∆Start range from 0.012 to 0.072 s depending on the method and from 0.012 to 0.112 s for ∆End. These values are in accordance with several recent publications on the topic (≥ 2019): Caramia et al [74] (∆Start = 0.022 s, ∆End = 0.024 s), Kidzinski et al [35] (∆Start = 0.010 s, ∆End = 0.013 s), Gadaleta et al [34] (∆Start ≈ ∆End ≈ 0.040 s) and Mei et al [75] (∆Start ≈ ∆End ≈ 0.020 s). These values (summarised in Table 2) are to be compared with the results presented in this section.…”

“…This gait pattern recognition can be used to successfully solve tasks for classifying gait disorders [29][30][31][32] or for extracting gait characteristics [33]. Deep learning methods have also been used to characterise the gait phase the subject is in, thus resulting in IC/FC detection from multiple accelerometers [34], 3D markers [35,36] or instrumented shoes [37].…”

A Review of Gait Phase Detection Algorithms for Lower Limb Prostheses

“…Firstly, the domain based on the threshold method [ 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 ], time-frequency analysis [ 18 , 19 , 20 , 21 ], and peak heuristic algorithms [ 16 , 19 , 22 , 23 , 24 , 25 ], which are also variations of the threshold method. Secondly, Machine Learning (ML) approaches are now among the most popular techniques to detect phases and events with various models such as Hidden Markov Models (HMM) [ 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 ], or several of the latest studies published based on the Artificial Neural Network technique (ANN) [ 35 , 36 , 37 , 38 ], Deep Learning Neural Network (DLNN) [ 39 , 40 , 41 , 42 , 43 ], a Convolutional Neural Network (CNN) [ 44 , 45 , 46 ], or [ 28 ] proposed a hybrid method that combined HMM and Fully connected Neural Networks (FNN). Different computation methodologies provide different performances regarding the parameters such as the number of detectable phases, events, and detection delay, which will be discussed in the next section.…”

Hidden Markov Model based stride segmentation on unsupervised free-living gait data in Parkinson’s disease patients

“…However, independent of the recording environment, a robust segmentation of individual strides from the continuous sensor data is one of the first steps in most wearable gait analysis systems and a crucial part of the underlying signal processing pipeline [7]. Various different approaches have been proposed in the literature to solve the problem of stride segmentation for clinical gait analysis applications [8][9][10][11][12][13][14][15][16][17][18][19][20][21][22]. Methods vary with sensor location, ranging from the upper body like wrist, chest or lower back [9,10] to the lower body with sensors attached to ankles or feet [11][12][13][14][15][17][18][19][20][21][22] as well as with sensor modalities like IMUs or pressure sensors [7].…”