Efficiency of deep neural networks for joint angle modeling in digital gait assessment

AbstractReliability and user compliance of the applied sensor system are two key issues of digital healthcare and biomedical informatics. For gait assessment applications, accurate joint angle measurements are important. Inertial measurement units (IMUs) have been used in a variety of applications and can also provide significant information on gait kinematics. However, the nonlinear mechanism of human locomotion results in moderate estimation accuracy of the gait kinematics and thus joint angles. To develop “digital twins” as a digital counterpart of body lower limb joint angles, three-dimensional gait kinematic data were collected. This work investigates the estimation accuracy of different neural networks in modeling lower body joint angles in the sagittal plane using the kinematic records of a single IMU attached to the foot. The evaluation results based on the root mean square error (RMSE) show that long short-term memory (LSTM) networks deliver superior performance in nonlinear modeling of the lower limb joint angles compared to other machine learning (ML) approaches. Accordingly, deep learning based on the LSTM architecture is a promising approach in modeling of gait kinematics using a single IMU, and thus can reduce the required physical IMUs attached on the subject and improve the practical application of the sensor system.

Download Full-text

Joint trajectory generator for powered orthosis based on gait modelling using PCA and FFT

Robotica ◽

10.1017/s0263574717000467 ◽

2017 ◽

Vol 36 (3) ◽

pp. 395-407 ◽

Cited By ~ 3

Author(s):

Nicholas B. Melo ◽

Carlos E. T. Dórea ◽

Pablo J. Alsina ◽

Márcio V. Araújo

Keyword(s):

Principal Component Analysis ◽

Fourier Series ◽

Energy Consumption ◽

Lower Limb ◽

Joint Angle ◽

Principal Component ◽

Orthotic Devices ◽

Joint Angles ◽

Anthropometric Features ◽

Trajectory Generator

SUMMARYIn this work, we propose a method able to find user-oriented gait trajectories that can be used in powered lower limb orthosis applications. Most research related to active orthotic devices focuses on solving hardware issues. However, the problem of generating a set of joint trajectories that are user-oriented still persists. The proposed method uses principal component analysis to extract shared features from a gait dataset, taking into consideration gait-related variables such as joint angle information and the user's anthropometric features, used directly in an orthosis application. The trajectories of joint angles used by the model are represented by a given number of harmonics according to their respective Fourier series analyses. This representation allows better performance of the model, whose capability to generate gait information is validated through experiments using a real active orthotic device, analysing both joint motor energy consumption and user metabolic effort.

Download Full-text

Adaptive Parameter Estimation of IIR System-Based WSN Using Multihop Diffusion in Distributed Approach

International Journal of Cognitive Informatics and Natural Intelligence ◽

10.4018/ijcini.2020100102 ◽

2020 ◽

Vol 14 (4) ◽

pp. 30-41

Author(s):

Meera Dash ◽

Trilochan Panigrahi ◽

Renu Sharma ◽

Mihir Narayan Mohanty

Keyword(s):

Impulse Response ◽

Sensor Node ◽

Finite Impulse Response ◽

Lms Algorithm ◽

Sensor System ◽

Superior Performance ◽

Estimation Accuracy ◽

Infinite Impulse Response ◽

Communication Overhead ◽

Least Mean Square

Distributed estimation of parameters in wireless sensor networks is taken into consideration to reduce the communication overhead of the network which makes the sensor system energy efficient. Most of the distributed approaches in literature, the sensor system is modeled with finite impulse response as it is inherently stable. Whereas in real time applications of WSN like target tracking, fast rerouting requires, infinite impulse response system (IIR) is used to model and that has been chosen in this work. It is assumed that every sensor node is equipped with IIR adaptive system. The diffusion least mean square (DLMS) algorithm is used to estimate the parameters of the IIR system where each node in the network cooperates themselves. In a sparse WSN, the performance of a DLMS algorithm reduces as the degree of the node decreases. In order to increase the estimation accuracy with a smaller number of iterations, the sensor node needs to share their information with more neighbors. This is feasible by communicating each node with multi-hop nodes instead of one-hop only. Therefore the parameters of an IIR system is estimated in distributed sparse sensor network using multihop diffusion LMS algorithm. The simulation results exhibit superior performance of the multihop diffusion LMS over non-cooperative and conventional diffusion algorithms.

Download Full-text

Joint Angle, Range of Motion, Force, and Moment Assessment: Responses of the Lower Limb to Ankle Plantarflexion and Dorsiflexion

Applied Bionics and Biomechanics ◽

10.1155/2021/1232468 ◽

2021 ◽

Vol 2021 ◽

pp. 1-13

Author(s):

Ukadike Chris Ugbolue ◽

Chloe Robson ◽

Emma Donald ◽

Kerry L. Speirs ◽

Frédéric Dutheil ◽

...

Keyword(s):

Range Of Motion ◽

Lower Limb ◽

Joint Angle ◽

Joint Angles ◽

Hip Joints ◽

Exercise Interventions ◽

Biomechanical Assessment ◽

Lower Limb Exercise ◽

Males And Females ◽

Foot Position

There is limited research on the biomechanical assessment of the lower limb joints in relation to dynamic movements that occur at the hip, knee, and ankle joints when performing dorsiflexion (DF) and plantarflexion (PF) among males and females. This study investigated the differences in joint angles (including range of motion (ROM)) and forces (including moments) between the left and right limbs at the ankle, knee, and hip joints during dynamic DF and PF movements in both males and females. Using a general linear model employing multivariate analysis in relation to the joint angle, ROM, force, and moment datasets, the results revealed significant main effects for gender, sidedness, phases, and foot position with respect to joint angles. Weak correlations were observed between measured biomechanical variables. These results provide insightful information for clinicians and biomechanists that relate to lower limb exercise interventions and modelling efficacy standpoints.

Download Full-text

Artificial Neural Networks in Motion Analysis—Applications of Unsupervised and Heuristic Feature Selection Techniques

Sensors ◽

10.3390/s20164581 ◽

2020 ◽

Vol 20 (16) ◽

pp. 4581

Author(s):

Marion Mundt ◽

Arnd Koeppe ◽

Franz Bamer ◽

Sina David ◽

Bernd Markert

Keyword(s):

Neural Networks ◽

Principal Component Analysis ◽

Motion Analysis ◽

Lower Limb ◽

Short Term Memory ◽

Principal Component ◽

Joint Angles ◽

Short Term ◽

Term Memory ◽

Long Short Term Memory

The use of machine learning to estimate joint angles from inertial sensors is a promising approach to in-field motion analysis. In this context, the simplification of the measurements by using a small number of sensors is of great interest. Neural networks have the opportunity to estimate joint angles from a sparse dataset, which enables the reduction of sensors necessary for the determination of all three-dimensional lower limb joint angles. Additionally, the dimensions of the problem can be simplified using principal component analysis. Training a long short-term memory neural network on the prediction of 3D lower limb joint angles based on inertial data showed that three sensors placed on the pelvis and both shanks are sufficient. The application of principal component analysis to the data of five sensors did not reveal improved results. The use of longer motion sequences compared to time-normalised gait cycles seems to be advantageous for the prediction accuracy, which bridges the gap to real-time applications of long short-term memory neural networks in the future.

Download Full-text

A Mechanical Sensor Designed for Dynamic Joint Angle Measurement

Journal of Healthcare Engineering ◽

10.1155/2017/8465212 ◽

2017 ◽

Vol 2017 ◽

pp. 1-12 ◽

Cited By ~ 1

Author(s):

Congo Tak-Shing Ching ◽

Su-Yu Liao ◽

Teng-Yun Cheng ◽

Chih-Hsiu Cheng ◽

Tai-Ping Sun ◽

...

Keyword(s):

Gait Analysis ◽

Linear Correlation ◽

Lower Limb ◽

Joint Angle ◽

Intraclass Correlation ◽

Reliability And Validity ◽

Angle Measurement ◽

Joint Angles ◽

Rehabilitation Programs ◽

Analysis System

Background. The measurement of the functional range of motion (FROM) of lower limb joints is an essential parameter for gait analysis especially in evaluating rehabilitation programs.Aim. To develop a simple, reliable, and affordable mechanical goniometer (MGR) for gait analysis, with six-degree freedom to dynamically assess lower limb joint angles.Design. Randomized control trials, in which a new MGR was developed for the measurements of FROM of lower limb joints.Setting. Reliability of the designed MGR was evaluated and validated by a motion analysis system (MAS).Population. Thirty healthy subjects participated in this study.Methods. Reliability and validity of the new MGR were tested by intraclass correlation coefficient (ICC), Bland-Altman plots, and linear correlation analysis.Results. The MGR has good inter- and intrarater reliability and validity withICC≥0.93(for both). Moreover, measurements made by MGR and MAS were comparable and repeatable with each other, as confirmed by Bland-Altman plots. Furthermore, a very high degree of linear correlation (R≥0.92for all joint angle measurements) was found between the lower limb joint angles measured by MGR and MAS.Conclusion. A simple, reliable, and affordable MGR has been designed and developed to aid clinical assessment and treatment evaluation of gait disorders.

Download Full-text

Inter-session repeatability of markerless motion capture gait kinematics

10.1101/2020.06.23.155358 ◽

2020 ◽

Author(s):

Robert M. Kanko ◽

Elise Laende ◽

W. Scott Selbie ◽

Kevin J. Deluzio

Keyword(s):

Motion Capture ◽

Lower Limb ◽

Video Data ◽

Joint Angles ◽

Markerless Tracking ◽

Gait Kinematics ◽

Markerless Motion Capture ◽

Adult Participants ◽

Total Variability ◽

Over Ground Walking

AbstractThe clinical uptake and influence of gait analysis has been hindered by inherent flaws of marker-based motion capture systems, which have long been the standard method for the collection of gait data including kinematics. Markerless motion capture offers an alternative method for the collection of gait kinematics that presents several practical benefits over marker-based systems. This work aimed to determine the reliability of lower limb gait kinematics from video based markerless motion capture using an established experimental protocol for testing reliability. Eight healthy adult participants performed three sessions of five over-ground walking trials in their own self-selected clothing, separated by an average of 8.5 days, while eight synchronized and calibrated cameras recorded video. 3D pose estimates from the video data were used to compute lower limb joint angles. Inter-session variability, inter-trial variability, and the variability ratio were used to assess the reliability of the gait kinematics. Relative to repeatability studies based on marker-based motion capture, inter-trial variability was slightly greater than previously reported for some angles, with an average across all joint angles of 2.2°. Inter-session variability was smaller on average than previously reported, with an average across all joint angles of 2.5°. Variability ratios were all smaller than those previously reported with an average of 1.2, indicating that the multi-session protocol increased the total variability of joint angles by only 20% of the inter-trial variability. These results indicate that gait kinematics measured using markerless tracking were less affected by multi-session protocols compared to marker-based motion capture.

Download Full-text

Prediction of lower limb joint angles and moments during gait using artificial neural networks

Medical & Biological Engineering & Computing ◽

10.1007/s11517-019-02061-3 ◽

2019 ◽

Vol 58 (1) ◽

pp. 211-225 ◽

Cited By ~ 5

Author(s):

Marion Mundt ◽

Wolf Thomsen ◽

Tom Witter ◽

Arnd Koeppe ◽

Sina David ◽

...

Keyword(s):

Neural Networks ◽

Artificial Neural Networks ◽

Lower Limb ◽

Joint Angles ◽

Artificial Neural

Download Full-text

Prediction of Limb Joint Angles Based on Multi-Source Signals by GS-GRNN for Exoskeleton Wearer

Sensors ◽

10.3390/s20041104 ◽

2020 ◽

Vol 20 (4) ◽

pp. 1104

Author(s):

Hualong Xie ◽

Guanchao Li ◽

Xiaofei Zhao ◽

Fei Li

Keyword(s):

Neural Network ◽

Lower Limb ◽

Prediction Accuracy ◽

Joint Angle ◽

Golden Section ◽

Human Movement ◽

Lower Limbs ◽

General Regression Neural Network ◽

Joint Angles ◽

Prediction Result

To enable exoskeleton wearers to walk on level ground, estimation of lower limb movement is particularly indispensable. In fact, it allows the exoskeleton to follow the human movement in real time. In this paper, the general regression neural network optimized by golden section algorithm (GS-GRNN) is used to realize prediction of the human lower limb joint angle. The human body hip joint angle and the surface electromyographic (sEMG) signals of the thigh muscles are taken as the inputs of a neural network to predict joint angles of lower limbs. To improve the prediction accuracy in different gait phases, the plantar pressure signals are also added into the input. After that, the error between the prediction result and the actual data decreases significantly. Finally, compared with the prediction result of the BP neural network, GRNN shows splendid prediction performance for its less processing time and higher prediction accuracy.

Download Full-text

Research on Joint-Angle Prediction Based on Artificial Neural Network for Above-Knee Amputees

Sensors ◽

10.3390/s21217199 ◽

2021 ◽

Vol 21 (21) ◽

pp. 7199

Author(s):

Jianyu Yang ◽

Guanchao Li ◽

Xiaofei Zhao ◽

Hualong Xie

Keyword(s):

Neural Network ◽

Ankle Joint ◽

Lower Limb ◽

Plantar Pressure ◽

Joint Angle ◽

Golden Section ◽

Prediction Method ◽

Maximum Error ◽

Generalized Regression Neural Network ◽

Joint Angles

In the current study, our research group proposed an asymmetric lower extremity exoskeleton to enable above-knee amputees to walk with a load. Due to the absence of shank and foot, the knee and ankle joint at the amputation side of the exoskeleton lack tracking targets, so it is difficult to realize the function of assisted walking when going up and downstairs. Currently, the use of lower-limb electromyography to predict the angles of lower limb joints has achieved remarkable results. However, the prediction effect was poor when only using electromyography from the thigh. Therefore, this paper introduces hip-angle and plantar pressure signals for improving prediction effect and puts forward a joint prediction method of knee- and ankle-joint angles by electromyography of the thigh, hip-joint angle, and plantar pressure signals. The generalized regression neural network optimized by the golden section method is used to predict the joint angles. Finally, the parameters (the maximum error, the Root-Mean-Square error (RMSE), and correlation coefficient (γ)) were calculated to verify the feasibility of the prediction method.

Download Full-text

Genome-Wide Association Studies Based on Equine Joint Angle Measurements Reveal New QTL Affecting the Conformation of Horses

Genes ◽

10.3390/genes10050370 ◽

2019 ◽

Vol 10 (5) ◽

pp. 370 ◽

Cited By ~ 5

Author(s):

Annik Imogen Gmel ◽

Thomas Druml ◽

Rudolf von Niederhäusern ◽

Tosso Leeb ◽

Markus Neuditschko

Keyword(s):

Joint Angle ◽

Association Studies ◽

Genome Wide Association ◽

Cranial Morphology ◽

Genome Wide Association Studies ◽

Joint Angles ◽

Mineral Density ◽

Mixed Effect ◽

Stifle Joint ◽

Genome Wide

The evaluation of conformation traits is an important part of selection for breeding stallions and mares. Some of these judged conformation traits involve joint angles that are associated with performance, health, and longevity. To improve our understanding of the genetic background of joint angles in horses, we have objectively measured the angles of the poll, elbow, carpal, fetlock (front and hind), hip, stifle, and hock joints based on one photograph of each of the 300 Franches-Montagnes (FM) and 224 Lipizzan (LIP) horses. After quality control, genome-wide association studies (GWASs) for these traits were performed on 495 horses, using 374,070 genome-wide single nucleotide polymorphisms (SNPs) in a mixed-effect model. We identified two significant quantitative trait loci (QTL) for the poll angle on ECA28 (p = 1.36 × 10−7), 50 kb downstream of the ALX1 gene, involved in cranial morphology, and for the elbow joint on ECA29 (p = 1.69 × 10−7), 49 kb downstream of the RSU1 gene, and 75 kb upstream of the PTER gene. Both genes are associated with bone mineral density in humans. Furthermore, we identified other suggestive QTL associated with the stifle joint on ECA8 (p = 3.10 × 10−7); the poll on ECA1 (p = 6.83 × 10−7); the fetlock joint of the hind limb on ECA27 (p = 5.42 × 10−7); and the carpal joint angle on ECA3 (p = 6.24 × 10−7), ECA4 (p = 6.07 × 10−7), and ECA7 (p = 8.83 × 10−7). The application of angular measurements in genetic studies may increase our understanding of the underlying genetic effects of important traits in equine breeding.

Download Full-text