Improving Gait Analysis Techniques with Markerless Pose Estimation Based on Smartphone Location

IntroductionInertial measurement units have been proposed for automated pose estimationand exercise monitoring in clinical settings. However, many existing methodsassume an extensive calibration procedure, which may not be realizable inclinical practice. In this study, an inertial measurement unit-based poseestimation method using extended Kalman filter and kinematic chain modelingis adapted for lower body pose estimation during clinical mobility testssuch as the single leg squat, and the sensitivity to parameter calibrationis investigated.MethodsThe sensitivity of pose estimation accuracy to each of the kinematic modeland sensor placement parameters was analyzed. Sensitivity analysis resultssuggested that accurate extraction of inertial measurement unit orientationon the body is a key factor in improving the accuracy. Hence, a simplecalibration protocol was proposed to reach a better approximation forinertial measurement unit orientation.ResultsAfter applying the protocol, the ankle, knee, and hip joint angle errorsimproved to , and , without the need for any other calibration.ConclusionsOnly a small subset of kinematic and sensor parameters contributesignificantly to pose estimation accuracy when using body worn inertialsensors. A simple calibration procedure identifying the inertial measurementunit orientation on the body can provide good pose estimationperformance.

Gait dysfunction is common in many clinical populations and often has a profound and deleterious impact on independence and quality of life. Gait analysis is a foundational component of rehabilitation because it is critical to identify and understand the specific deficits that should be targeted prior to the initiation of treatment. Unfortunately, current state-of-the-art approaches to gait analysis (e.g., marker-based motion capture systems, instrumented gait mats) are largely inaccessible due to prohibitive costs of time, money, and effort required to perform the assessments. Here, we demonstrate the ability to perform quantitative gait analyses in multiple clinical populations using only simple videos recorded using low-cost devices (tablets). We report four primary advances: 1) a novel, versatile workflow that leverages an open-source human pose estimation algorithm (OpenPose) to perform gait analyses using videos recorded from multiple different perspectives (e.g., frontal, sagittal), 2) validation of this workflow in three different populations of participants (adults without gait impairment, persons post-stroke, and persons with Parkinson's disease) via comparison to ground-truth three-dimensional motion capture, 3) demonstration of the ability to capture clinically relevant, condition-specific gait parameters, and 4) tracking of within-participant changes in gait, as is required to measure progress in rehabilitation and recovery. Importantly, our workflow has been made freely available and does not require prior gait analysis expertise. The ability to perform quantitative gait analyses in nearly any setting using only low-cost devices and computer vision offers significant potential for dramatic improvement in the accessibility of clinical gait analysis across different patient populations.

Marker-less analysis of human gait has made considerable progress in recent years. However, developing a gait analysis system capable of extracting reliable and precise kinematic data in a standard and unobtrusive manner remains an open challenge. This narrative review considers the transformation of methods for extracting gait extremity information from videos or images, perceived how analysis methods have improved from arduous manual procedures to semi-objective and objective marker-based systems and then marker-less systems. The gait analysis systems widely used restrict the analysis process with the use of markers, inhibited environmental conditions, and long processing duration. Such limitations can impede the use of a gait analysis system in multiple applications. Advancement in marker-less pose estimation and Q-learning-based techniques are opening the possibility of adopting productive methods for estimating precise poses of humans and information of movement from video frames. Vision-Based gait analysis techniques are capable of providing a cost-effective, unobtrusive solution for estimation of stick images and thus the analysis of the gait. This work provides a comprehensive review of marker-less computer vision and deep neural network-based gait analysis, parameters, design specifications, and the latest trends. This survey provides a birds-eye view of the domain. This review aims to introduce the latest trends in gait analysis using computer vision methods thus provide a single platform to learn various marker-less methods for the analysis of the gait that is likely to have a future impact in bio-mechanics while considering the challenges with accuracy and robustness that are yet to be addressed.KeywordsHuman Gait analysisMarker-less techniquesPose estimationStick imagesAbnormality detection

During the lowering phase of a squat, it has been observed that a posterior pelvic tilt (PPT) may occur when squatting to full depth. Research suggests that defaulting to compensatory movement strategies, such as PPT, during the squat may correlate with risk of lower extremity and trunk pathology. The purpose of this study was to examine hip flexion (HF) angles at the point when PPT occurs among three conditions: standard squats, heel raise squats, and supine passive HF; analyzing the differences in depth between standard and heel raise squats; and calculating differences in knee angles and ankle excursion between standard and heel raise squats. 28 participants performed bodyweight squats and underwent supine passive HF while outfitted with 32 retroreflective motion capture markers. Hip, knee, and ankle joint angles were extracted at the point of PPT. A one-way repeated measures ANOVA was used to determine differences in hip joint angles between conditions, and a paired sample t-test was used to compare knee angles, ankle excursion, and squat depth between standard and heel raise squats. HF angles at PPT remained unchanged across all conditions. However, significantly greater knee flexion, ankle excursion, and squat depth were observed in the heel raise squats compared to the standard squats. Results suggest that PPT is a compensatory movement that occurs as the femur compresses into the acetabulum once hip flexion has been exhausted.

Analysis and evaluation of hemiplegic gait based on wearable sensor network

The effects of fatigue on the relationship between ankle angle at initial contact and the knee and hip joints in landing: Assessing the risk of ACL injury

This study aimed to investigate the effects of genu recurvatum, which is considered to carry a high risk for anterior cruciate ligament (ACL) injury, on healthy and post-ACL injury gait and lower extremity muscle strength. Subjects were 36 patients with ACL-deficient knee and 40 healthy controls without pain or restricted range of motion of the lower extremity during gait. The knee joints of all subjects were examined; those with over 10° hyperextension of both knees were defined as exhibiting genu recurvatum. On this basis, the subjects were further subdivided into two groups: with or without genu recurvatum. A three-dimensional motion analysis system and force plates were used for gait analysis. Isokinetic dynamometers were used to measure knee muscle strength. There were no differences in joint angles, joint moments, or components of ground reaction force during gait or in knee strength for the healthy control subjects with and without genu recurvatum. ACL-deficient subjects without genu recurvatum showed a decrease in knee angles during the stance phase and a decrease in extension moments during the early stance phase compared with ACL-deficient subjects with genu recurvatum and controls. In contrast, neither knee angles nor extension moments during the stance phase differed significantly between ACL-deficient subjects with genu recurvatum and controls. This study provides clinically relevant information regarding the effects of genu recurvatum on gait parameters. The results suggest that in ACL injuries, the presence of genu recurvatum alters gait pattern. Consideration of the presence of genu recurvatum would be useful during rehabilitation following ACL injuries or ACL reconstruction. LEVEL OF EVIDENCE: II. Prospective comparative study.

Variations in knee angles are important parameters used for clinical gait analysis. They are measured using gold-standard marker-based motion capture systems in clinical research. However, these systems are difficult to operate for longer durations and are limited to laboratory environments. Although camera-based approaches are available, they require a direct line of sight, adequate ambient lighting, and a clear background. Ultra-wideband radar technology (UWB) has the potential to address these limitations and support moving away from strict laboratory settings. This paper presents a UWB radar-based gait analysis system to capture walking gait, a novel algorithm, a unique detection intensity level algorithm (DILA), registering detections in a 3D space, and a PointNet-inspired neural network (DIL-PointNet) for knee angle estimation. The proposed system is validated against knee angles calculated with the Qualisys marker-based motion capture system. It achieved a mean absolute error (MAE) of 7.9° and a root mean squared error (RMSE) of 11.1°. Thus, the proposed system has the potential for long-term remote gait monitoring in real world environments.

We propose and demonstrate an efficient pose estimation method, specifically small tilt angle variation, for visual servo control system design using a monocular camera, eliminating the need for iterative computations. This paper addresses the challenges associated with accurately estimating the relative pose between a chaser, equipped with a monocular camera, and a target that provides visual cues, particularly in scenarios where the captured image shows only minimal variations in the target's tilt angle. Traditional approaches often rely on direct feedback methods from image data or designs for pose estimation and control separately, necessitating computationally expensive iterations for pose estimation or complex implementation of visual servoing. To address these challenges, we adopt an extended Kalman filter, known as a robust nonlinear observer, and propose a non‐iterative approach that integrates motion dynamics models and control inputs into pose estimation from image data. This approach enables efficient integration of pose estimation and control solely relying on a monocular camera, reducing computational costs and improving robustness. Experimental validation using the ball‐and‐plate system demonstrates the effectiveness of our proposed method. In our setup, the camera is deliberately positioned directly above the pose estimation target, creating a scenario with minimal image variation in the tilt angle. Despite this condition, our approach achieves efficient and accurate pose estimation, demonstrating its feasibility for high‐speed pose estimation and control without iterative computations.

Several reports that have compared the kinematic differences between treadmill (TM) and overground (OG) walking in healthy subjects. Studies on temporal aspects of gait in lean, healthy individuals typically report increased stride frequency and decreased stride length as the main difference between the two walking modalities. Findings are varied for differences in peak joint angles but trend towards minimal differences in peak values. Obese individuals display markedly different kinematics relative to lean individuals and are often not experienced TM walkers. The dimensions of standard laboratory TMs may also constrain limb excursions or elicit the need for increased stability. PURPOSE: To compare selected kinematic parameters of TM and OG walking in obese women. METHODS: 12 obese women (28.7 ± 9.1 years; 36.7 ± 6.4 kg·m-2; 45.6 ± 5.1% body fat) walked OG and on a TM at their preferred OG walking speed (1.37 ± 0.14 m·s-1). Kinematic measures on the TM were recorded after 6 minute of walking. Measured kinematic variables across 5 strides included stride length, stride frequency, stance time, and peak sagittal and frontal plane hip, knee, and ankle joint angles during the stance phase of the gait cycle. Paired T-tests were calculated to assess statistical significance (α = 0.05). RESULTS: Obese women had significantly greater stance time on the TM than during OG walking (p=0.01). Peak knee flexion during stance was greater on the TM (p<0.01) while ankle dorsiflexion during midstance (p=0.05) and ankle eversion range of motion (p<0.01) were significantly greater during OG walking. There were no significant differences between conditions for stride frequency, stride length, peak hip flexion, hip extension, hip range of motion, hip abduction, knee adduction, ankle plantar flexion during midstance, or peak ankle eversion during stance (p>0.05). CONCLUSIONS: Obese women display different kinematics on the TM relative to OG walking. In addition, the differences found in this study do not match reports from similar studies on lean individuals. The greater limb dimensions of the obese population potentially necessitate that different strategies be used for forward locomotion on a TM compared to OG. The results of this study indicate that several kinematic measures recorded on a TM do not replicate OG walking in obese adults.

For robot-assisted assembly of complex optical systems, the alignment is facilitated by an accurate pose estimation of its components. However, wavefront-based pose estimation is typically ill-conditioned due to the inherent geometry of conventional industrially manufactured optical components. Therefore, we propose a novel approach in this paper to increase wavefront-based pose estimation accuracy via the design of freeform optics. For this purpose, an optimization problem is derived that parameterizes the component's surfaces by a predetermined freeform surface model. To show the efficacy of our approach, we provide simulation results to compare the pose estimation accuracy for a variety of optical designs. As an application example for the resulting improved pose estimation, a hand-eye calibration of a wavefront sensor is performed. This calibration originates from the field of robotics and represents the identification of a sensor coordinate system with respect to a global reference frame. For quantitative evaluation, the calibrating results are first presented with the aid of simulation data. Finally, the practical feasibility is demonstrated using a conventional industrial robot and additively manufactured freeform lenses.

Joint object recognition and pose estimation using multiple-anchor triplet learning of canonical plane

Gait analysis is an important research field in neurodegenerative disease diagnosis. Various IMU-based and vision-based methods have been developed for measuring joint angles during walking process. Dynamic Vision Sensor (DVS) is a neuromorphic engineering product which outputs sequences of events describing the level of brightness change at the pixel level in an asynchronized manner, instead of discrete frames at a predefined frame rate. Due to its microsecond level resolution, DVS is regarded possessing a huge potential to dramatically increase the speed of sensing pipeline for various applications which traditional CMOS camera cannot fulfill. Hence, in this paper we introduce a novel system for gait analysis application; DVS and special markers are used to detect the ankle joint during walking and a method is developed to calculate the desired angle. Further, the robust locally weighted regression is employed as signal smoothing method to reduce the amount of noise. In order to evaluation its performance, an Inertial Measurement Unit (IMU)-based sensory system is also examined and compared in the same experiment and goniometer is used for providing the ground truth. By comparing the captured ankle joint angle trajectories using Dynamic Time Warping (DTW), DVS-based system appears to have higher accuracy in terms of ankle joint angle detection than IMU-based system.

Knee (femorotibial), ankle (talocrural) and metatarsophalangeal joint angles during posture and locomotion were determined for the adult Striped skunk, Mephitis mephitis. Cinematography of animals moving overground and cineradiography of animals walking on a treadmill are the basis for joint angle measurements. These data allowed for the documentation of muscle length change of the primary ankle extensors, soleus, medial and lateral gastrocnemius and plantaris during stepping. Whole muscle length—active tension curves were determined on anaesthetized skunks. Appropriate moment arms were measured over the physiological range of knee and ankle angles. This study provides some mechanical and musculoskeletal data of the ankle of M. mephitis and provides for comparison and contrast to similar data for the domestic cat. Major findings include: (1) the length—active tetanic tension curves of the ankle extensors tend to be broad over a wider range of lengths in M. mephitis than in the cat, (2) the range of lengths at which the ankle extensors are actively used is relatively narrow in both species, (3) as both animals walk faster the working range of the muscles is closer to the maximum in situ length, (4) a broader range of ankle joint angles exists for M. rnephitis than for the cat at which the length—active tension curve and moment arm curve interact to provide maximum or near maximum torque, and (5) in general, the stance phase of the skunk is initiated at muscle lengths (relative to standing length) that are much shorter than those of the cat which may result in greater elastic storage for the cat durine E2.These data coupled with an analysis of gait selection of M. mephitis lead us to conclude that many features of the posture, locomotion and musculoskeletal design of the ankle extensor system of this species are primitive. The cat deviates from the primitive condition to some extent related to its more extreme specialization toward a jumping and sprinting way of life. Nevertheless, the domestic cat is a useful musculoskeletal model for studies concerned with the neural control of mammalian locomotion.

The ability to accurately and non-invasively measure 3D mass centre positions and their derivatives can provide rich insight into the physical demands of sports training and competition. This study examines a method for non-invasively measuring mass centre velocities using markerless human pose estimation and Kalman smoothing. Marker (Qualysis) and markerless (OpenPose) motion capture data were captured synchronously for sprinting and skeleton push starts. Mass centre positions and velocities derived from raw markerless pose estimation data contained large errors for both sprinting and skeleton pushing (mean ± SD = 0.127 ± 0.943 and −0.197 ± 1.549 m·s−1, respectively). Signal processing methods such as Kalman smoothing substantially reduced the mean error (±SD) in horizontal mass centre velocities (0.041 ± 0.257 m·s−1) during sprinting but the precision remained poor. Applying pose estimation to activities which exhibit unusual body poses (e.g., skeleton pushing) appears to elicit more erroneous results due to poor performance of the pose estimation algorithm. Researchers and practitioners should apply these methods with caution to activities beyond sprinting as pose estimation algorithms may not generalise well to the activity of interest. Retraining the model using activity specific data to produce more specialised networks is therefore recommended.