The efficacy of a video-based marker-less tracking system for gait analysis

Most clinical gait analyses are conducted using motion capture systems which track retro-reflective markers that are placed on key landmarks of the participants. An alternative to a three-dimensional (3D) motion capture, marker-based, optical camera system may be a marker-less video-based tracking system. The aim of our study was to investigate the efficacy of the use of a marker-less tracking system in the calculation of 3D joint angles for possible use in clinical gait analysis. Ten participants walked and jogged on a treadmill and their kinematic data were captured with a marker and marker-less tracking system simultaneously. The hip, knee and ankle angles in the frontal, sagittal and transverse planes were computed. Root Mean Square differences (RMS_diff) between corresponding angles for each participant’s support phase were calculated and averaged to derive the mean within-subject RMS_diff. These within-subject means were averaged to obtain the mean between-subject RMS_diff for the relevant joint angles in the two gait conditions (walking and jogging). The RMS_diff between the two tracking systems was less than 1° for all rotations of the three joint angles of the hip and knee. However, there were slightly larger differences in the ankle joint angles. The results of this study suggest a potential application in gait analysis in clinical settings where observations of anatomical motions may provide meaningful feedback.     

Measurement of multi-segment foot joint angles during gait using a wearable system

Usually the measurement of multi-segment foot and ankle complex kinematics is done with stationary motion capture devices which are limited to use in a gait laboratory. This study aimed to propose and validate a wearable system to measure the foot and ankle complex joint angles during gait in daily conditions, and then to investigate its suitability for clinical evaluations. The foot and ankle complex consisted of four segments (shank, hindfoot, forefoot, and toes), with an inertial measurement unit (3D gyroscopes and 3D accelerometers) attached to each segment. The angles between the four segments were calculated in the sagittal, coronal, and transverse planes using a new algorithm combining strap-down integration and detection of low-acceleration instants. To validate the joint angles measured by the wearable system, three subjects walked on a treadmill for five minutes at three different speeds. A camera-based stationary system that used a cluster of markers on each segment was used as a reference. To test the suitability of the system for clinical evaluation, the joint angle ranges were compared between a group of 10 healthy subjects and a group of 12 patients with ankle osteoarthritis, during two 50-m walking trials where the wearable system was attached to each subject. On average, over all joints and walking speeds, the RMS differences and correlation coefficients between the angular curves obtained using the wearable system and the stationary system were 1 deg and 0.93, respectively. Moreover, this system was able to detect significant alteration of foot and ankle function between the group of patients with ankle osteoarthritis and the group of healthy subjects. In conclusion, this wearable system was accurate and suitable for clinical evaluation when used to measure the multi-segment foot and ankle complex kinematics during long-distance walks in daily life conditions.     

Accuracy and repeatability of joint angles measured using a single camera markerless motion capture system

Markerless motion capture systems have developed in an effort to evaluate human movement in a natural setting. However, the accuracy and reliability of these systems remain understudied. Therefore, the goals of this study were to quantify the accuracy and repeatability of joint angles using a single camera markerless motion capture system and to compare the markerless system performance with that of a marker-based system. A jig was placed in multiple static postures with marker trajectories collected using a ten camera motion analysis system. Depth and color image data were simultaneously collected from a single Microsoft Kinect camera, which was subsequently used to calculate virtual marker trajectories. A digital inclinometer provided a measure of ground-truth for sagittal and frontal plane joint angles. Joint angles were calculated with marker data from both motion capture systems using successive body-fixed rotations. The sagittal and frontal plane joint angles calculated from the marker-based and markerless system agreed with inclinometer measurements by <0.5°. The systems agreed with each other by <0.5° for sagittal and frontal plane joint angles and <2° for transverse plane rotation. Both systems showed a coefficient of reliability <0.5° for all angles. These results illustrate the feasibility of a single camera markerless motion capture system to accurately measure lower extremity kinematics and provide a first step in using this technology to discern clinically relevant differences in the joint kinematics of patient populations.     

Assessment of the measurement accuracy of inertial sensors during different tasks of daily living

The low cost and ease of use of inertial measurement units (IMUs) make them an attractive option for motion analysis tasks that cannot be easily measured in a laboratory. To date, only a limited amount of research has been conducted comparing commercial IMU systems to optoelectronic systems, the gold standard, for everyday tasks like stair climbing and inclined walking. In this paper, the 3D joint angles of the lower limbs are determined using both an IMU system and an optoelectronic system for twelve participants during stair ascent and descent, and inclined, declined and level walking. Three different datasets based on different hardware and anatomical models were collected for the same movement in an effort to determine the cause and quantify the errors involved with the analysis. Firstly, to calculate software errors, two different anatomical models were compared for one hardware system. Secondly, to calculate hardware errors, results were compared between two different measurement systems using the same anatomical model. Finally, the overall error between both systems with their native anatomical models was calculated. Statistical analysis was performed using statistical parametric mapping. When both systems were evaluated based on the same anatomical model, the number of trials with significant differences decreased markedly. Thus, the differences in joint angle measurement can mainly be attributed to the variability in the anatomical models used for calculations and not to the IMU hardware.     

Physical sensor difference-based method and virtual sensor difference-based method for visual and quantitative estimation of lower limb 3D gait posture using accelerometers and magnetometers

An approach using a physical sensor difference-based algorithm and a virtual sensor difference-based algorithm to visually and quantitatively confirm lower limb posture was proposed. Three accelerometers and two MAG(3)s (inertial sensor module) were used to measure the accelerations and magnetic field data for the calculation of flexion/extension (FE) and abduction/adduction (AA) angles of hip joint and FE, AA and internal/external rotation (IE) angles of knee joint; then, the trajectories of knee and ankle joints were obtained with the joint angles and segment lengths. There was no integration of acceleration or angular velocity for the joint rotations and positions, which is an improvement on the previous method in recent literature. Compared with the camera motion capture system, the correlation coefficients in five trials were above 0.91 and 0.92 for the hip FE and AA, respectively, and higher than 0.94, 0.93 and 0.93 for the knee joint FE, AA and IE, respectively.     

Gait posture estimation using wearable acceleration and gyro sensors

A method for gait analysis using wearable acceleration sensors and gyro sensors is proposed in this work. The volunteers wore sensor units that included a tri-axis acceleration sensor and three single axis gyro sensors. The angular velocity data measured by the gyro sensors were used to estimate the translational acceleration in the gait analysis. The translational acceleration was then subtracted from the acceleration sensor measurements to obtain the gravitational acceleration, giving the orientation of the lower limb segments. Segment orientation along with body measurements were used to obtain the positions of hip, knee, and ankle joints to create stick figure models of the volunteers. This method can measure the three-dimensional positions of joint centers of the hip, knee, and ankle during movement. Experiments were carried out on the normal gait of three healthy volunteers. As a result, the flexion–extension (F–E) and the adduction–abduction (A–A) joint angles of the hips and the flexion–extension (F–E) joint angles of the knees were calculated and compared with a camera motion capture system. The correlation coefficients were above 0.88 for the hip F–E, higher than 0.72 for the hip A–A, better than 0.92 for the knee F–E. A moving stick figure model of each volunteer was created to visually confirm the walking posture. Further, the knee and ankle joint trajectories in the horizontal plane showed that the left and right legs were bilaterally symmetric.     

Physical sensor difference-based method and virtual sensor difference-based method for visual and quantitative estimation of lower limb 3D gait posture using accelerometers and magnetometers

An approach using a physical sensor difference-based algorithm and a virtual sensor difference-based algorithm to visually and quantitatively confirm lower limb posture was proposed. Three accelerometers and two MAG³s (inertial sensor module) were used to measure the accelerations and magnetic field data for the calculation of flexion/extension (FE) and abduction/adduction (AA) angles of hip joint and FE, AA and internal/external rotation (IE) angles of knee joint; then, the trajectories of knee and ankle joints were obtained with the joint angles and segment lengths. There was no integration of acceleration or angular velocity for the joint rotations and positions, which is an improvement on the previous method in recent literature. Compared with the camera motion capture system, the correlation coefficients in five trials were above 0.91 and 0.92 for the hip FE and AA, respectively, and higher than 0.94, 0.93 and 0.93 for the knee joint FE, AA and IE, respectively.     

Design and validation of surface-marker clusters for the quantification of joint rotations in general movements in early infancy

Lack of complexity in general movements in early infancy is an important marker of potential motor disorders of neurological origin, such as cerebral palsy. Quantitative approaches to characterising this complexity are hampered by experimental difficulties in recording from infants in their first few months of life. The aim of this study was to design and validate bespoke surface-marker clusters to facilitate data acquisition and enable full quantification of joint rotations. The clusters were validated by recording the controlled movements of a soft-body dummy doll simultaneously with an optical (Qualisys) and inertial (XSens) motion capture system. The angles estimated from the optical system were compared with those measured by the inertial system. We demonstrate that the surface-marker based approach compares well with the use of an inertial system to obtain "direct" readings of the rotations whilst alleviating the issues associated with the use of an optical motion capture system. We briefly report use of this technique in 1-5 month old infants. By enabling full quantification of joint rotation, use of the custom made markers could pave the way for early diagnosis of movement disorders.     

Identification of passive elastic joint moment-angle relationships in the lower extremity

The purpose of this study was to develop a method for identifying subject-specific passive elastic joint moment-angle relationships in the lower extremity, which could subsequently be used to estimate passive contributions to joint kinetics during gait. Twenty healthy young adults participated in the study. Subjects were positioned side-lying with their dominant limb supported on a table via low-friction carts. A physical therapist slowly manipulated the limb through full sagittal hip, knee, and ankle ranges of motion using two hand-held 3D load cells. Lower extremity kinematics, measured with a passive marker motion capture system, and load cell readings were used to compute joint angles and associated passive joint moments. We formulated a passive joint moment-angle model that included eight exponential functions to account for forces generated via the passive stretch of uni-articular structures and bi-articular muscles. Model parameters were estimated for individual subjects by minimizing the sum of squared errors between model predicted and experimentally measured moments. The model predictions closely replicated measured joint moments with average root-mean-squared errors of 2.5, 1.4, and 0.7 Nm about the hip, knee, and ankle respectively. We show that the models can be coupled with gait kinematics to estimate passive joint moments during walking. Passive hip moments were substantial from terminal stance through initial swing, with energy being stored as the hip extended and subsequently returned during pre- and initial swing. We conclude that the proposed methodology could provide quantitative insights into the potentially important role that passive mechanisms play in both normal and abnormal gait.     

The effects of alignment of an articulated ankle-foot orthosis on lower limb joint kinematics and kinetics during gait in individuals post-stroke

Mechanical tuning of an ankle-foot orthosis (AFO) is important in improving gait in individuals post-stroke. Alignment and resistance are two factors that are tunable in articulated AFOs. The aim of this study was to investigate the effects of changing AFO ankle alignment on lower limb joint kinematics and kinetics with constant dorsiflexion and plantarflexion resistance in individuals post-stroke. Gait analysis was performed on 10 individuals post-stroke under four distinct alignment conditions using an articulated AFO with an ankle joint whose alignment is adjustable in the sagittal plane. Kinematic and kinetic data of lower limb joints were recorded using a Vicon 3-dimensional motion capture system and Bertec split-belt instrumented treadmill. The incremental changes in the alignment of the articulated AFO toward dorsiflexion angles significantly affected ankle and knee joint angles and knee joint moments while walking in individuals post-stroke. No significant differences were found in the hip joint parameters. The alignment of the articulated AFO was suggested to play an important role in improving knee joint kinematics and kinetics in stance through improvement of ankle joint kinematics while walking in individuals post-stroke. Future studies should investigate long-term effects of AFO alignment on gait in the community in individuals post-stroke.     

Lower-limb joint work and power are modulated during load carriage based on load configuration and walking speed

Soldiers regularly transport loads weighing >20 kg at slow speeds for long durations. These tasks elicit high energetic costs through increased positive work generated by knee and ankle muscles, which may increase risk of muscular fatigue and decrease combat readiness. This study aimed to determine how modifying where load is borne changes lower-limb joint mechanical work production, and if load magnitude and/or walking speed also affect work production. Twenty Australian soldiers participated, donning a total of 12 body armor variations: six different body armor systems (one standard-issue, two commercially available [cARM1-2], and three prototypes [pARM1-3]), each worn with two different load magnitudes (15 and 30 kg). For each armor variation, participants completed treadmill walking at two speeds (1.51 and 1.83 m/s). Three-dimensional motion capture and force plate data were acquired and used to estimate joint angles and moments from inverse kinematics and dynamics, respectively. Subsequently, hip, knee, and ankle joint work and power were computed and compared between armor types and walking speeds. Positive joint work over the stance phase significantly increased with walking speed and carried load, accompanied by 2.3–2.6% shifts in total positive work production from the ankle to the hip (p < 0.05). Compared to using cARM1 with 15 kg carried load, carrying 30 kg resulted in significantly greater hip contribution to total lower-limb positive work, while knee and ankle work decreased. Substantial increases in hip joint contributions to total lower-limb positive work that occur with increases in walking speed and load magnitude highlight the importance of hip musculature to load carriage walking.     

Ambulatory measurement of 3D knee joint angle

Three-dimensional measurement of joint motion is a promising tool for clinical evaluation and therapeutic treatment comparisons. Although many devices exist for joints kinematics assessment, there is a need for a system that could be used in routine practice. Such a system should be accurate, ambulatory, and easy to use. The combination of gyroscopes and accelerometers (i.e., inertial measurement unit) has proven to be suitable for unrestrained measurement of orientation during a short period of time (i.e., few minutes). However, due to their inability to detect horizontal reference, inertial-based systems generally fail to measure differential orientation, a prerequisite for computing the three-dimentional knee joint angle recommended by the Internal Society of Biomechanics (ISB). A simple method based on a leg movement is proposed here to align two inertial measurement units fixed on the thigh and shank segments. Based on the combination of the former alignment and a fusion algorithm, the three-dimensional knee joint angle is measured and compared with a magnetic motion capture system during walking. The proposed system is suitable to measure the absolute knee flexion/extension and abduction/adduction angles with mean (SD) offset errors of -1 degree (1 degree ) and 0 degrees (0.6 degrees ) and mean (SD) root mean square (RMS) errors of 1.5 degrees (0.4 degrees ) and 1.7 degrees (0.5 degrees ). The system is also suitable for the relative measurement of knee internal/external rotation (mean (SD) offset error of 3.4 degrees (2.7 degrees )) with a mean (SD) RMS error of 1.6 degrees (0.5 degrees ). The method described in this paper can be easily adapted in order to measure other joint angular displacements such as elbow or ankle.     

Comparison of strain-gage and fiber-optic goniometry for measuring knee kinematics during activities of daily living and exercise

There is increasing interest in wearable sensor technology as a tool for rehabilitation applications in community or home environments. Recent studies have focused on evaluating inertial based sensing (accelerometers, gyroscopes, etc.) that provide only indirect measures of joint motion. Measurement of joint kinematics using flexible goniometry is more direct, and still popular in laboratory environments, but has received little attention as a potential tool for wearable systems. The aim of this study was to compare two goniometric devices: a traditional strain-gauge flexible goniometer, and a fiberoptic flexible goniometer, for measuring dynamic knee flexion/extension angles during activity of daily living: chair rise, and gait; and exercise: deep knee bends, against joint angles computed from a "gold standard" Vicon motion tracking system. Six young adults were recruited to perform the above activities in the lab while wearing a goniometer on each knee, and reflective markers for motion tracking. Kinematic data were collected simultaneously from the goniometers (one on each leg) and the motion tracking system (both legs). The results indicate that both goniometers were within 2-5 degrees of the Vicon angles for gait and chair rise. For some deep knee bend trials, disagreement with Vicon angles exceeded ten degrees for both devices. We conclude that both goniometers can record ADL knee movement faithfully and accurately, but should be carefully considered when high (>120?deg) knee flexion angles are required.     

Disability and Fatigue Can Be Objectively Measured in Multiple Sclerosis

Background

The available clinical outcome measures of disability in multiple sclerosis are not adequately responsive or sensitive.

Objective

To investigate the feasibility of inertial sensor-based gait analysis in multiple sclerosis.

Methods

A cross-sectional study of 80 multiple sclerosis patients and 50 healthy controls was performed. Lower-limb kinematics was evaluated by using a commercially available magnetic inertial measurement unit system. Mean and standard deviation of range of motion (mROM, sROM) for each joint of lower limbs were calculated in one minute walking test. A motor performance index (E) defined as the sum of sROMs was proposed.

Results

We established two novel observer-independent measures of disability. Hip mROM was extremely sensitive in measuring lower limb motor impairment, being correlated with muscle strength and also altered in patients without clinically detectable disability. On the other hand, E index discriminated patients according to disability, being altered only in patients with moderate and severe disability, regardless of walking speed. It was strongly correlated with fatigue and patient-perceived health status.

Conclusions

Inertial sensor-based gait analysis is feasible and can detect clinical and subclinical disability in multiple sclerosis.     

Segment and joint angles of hind limb during bipedal and quadrupedal walking of the bonobo (Pan paniscus)

We describe segment angles (trunk, thigh, shank, and foot) and joint angles (hip, knee, and ankle) for the hind limbs of bonobos walking bipedally ("bent-hip bent-knee walking," 17 sequences) and quadrupedally (33 sequences). Data were based on video recordings (50 Hz) of nine subjects in a lateral view, walking at voluntary speed. The major differences between bipedal and quadrupedal walking are found in the trunk, thigh, and hip angles. During bipedal walking, the trunk is approximately 33-41 degrees more erect than during quadrupedal locomotion, although it is considerably more bent forward than in normal human locomotion. Moreover, during bipedal walking, the hip has a smaller range of motion (by 12 degrees ) and is more extended (by 20-35 degrees ) than during quadrupedal walking. In general, angle profiles in bonobos are much more variable than in humans. Intralimb phase relationships of subsequent joint angles show that hip-knee coordination is similar for bipedal and quadrupedal walking, and resembles the human pattern. The coordination between knee and ankle differs much more from the human pattern. Based on joint angles observed throughout stance phase and on the estimation of functional leg length, an efficient inverted pendulum mechanism is not expected in bonobos.     

Estimation of minimum ground clearance (MGC) using body-worn inertial sensors

Objective assessment of balance and mobility in elderly populations using body-worn sensors has recently become a prevalent theme in falls-related research. Recent research by the authors identified mean absolute-valued vertical angular velocity measured using shank mounted inertial sensors during a timed-up-and-go test as having a strong association with falls history in a group of elderly adults. This study aimed to investigate the clinical relevance of this parameter by exploring the relationship between it and minimum ground clearance (MGC) measured with an optical motion capture system. MGC is an important variable when considering trip-related falls risk. This paper also presents a method of estimating properties of MGC during walking, across a range of speeds and gait patterns, using body-worn inertial sensors. We found that mean MGC and coefficient of variation (CV) MGC are correlated with mean absolute-valued vertical angular velocity and acceleration as measured by shank or foot mounted inertial sensors. Regression models generated using inertial sensor derived variables were used to robustly estimate the mean MGC and CV MGC measured by an optical marker-tracking system. Foot-mounted sensors were found to yield slightly better results than sensors on the shank. Different walking speeds and gait patterns were not found to influence the accuracy of the models. We conclude that these findings have the potential to evaluate a walking trial using body-worn inertial sensors, which could then be used to identify individuals with increased risk of unprovoked collisions with the ground during locomotion.     

Oscillation and Reaction Board Techniques for Estimating Inertial Properties of a Below-knee Prosthesis

The purpose of this study was two-fold: 1) demonstrate a technique that can be used to directly estimate the inertial properties of a below-knee prosthesis, and 2) contrast the effects of the proposed technique and that of using intact limb inertial properties on joint kinetic estimates during walking in unilateral, transtibial amputees. An oscillation and reaction board system was validated and shown to be reliable when measuring inertial properties of known geometrical solids. When direct measurements of inertial properties of the prosthesis were used in inverse dynamics modeling of the lower extremity compared with inertial estimates based on an intact shank and foot, joint kinetics at the hip and knee were significantly lower during the swing phase of walking. Differences in joint kinetics during stance, however, were smaller than those observed during swing. Therefore, researchers focusing on the swing phase of walking should consider the impact of prosthesis inertia property estimates on study outcomes. For stance, either one of the two inertial models investigated in our study would likely lead to similar outcomes with an inverse dynamics assessment.     

The effect of direct measurement versus cadaver estimates of anthropometry in the calculation of joint moments during above-knee prosthetic gait in pediatrics

Joint reaction forces, moments and powers are important in interpreting gait mechanics and compensatory strategies used by patients walking with above-knee prostheses. Segmental anthropometrics, required to calculate joint moments, are often estimated using data from cadaver studies. However, these values may not be accurate for patients following amputation as prostheses are composed of non-biologic material. The purpose of this study was to compare joint moments using anthropometrics calculated from cadaver studies versus direct measurements of the residual limb and prosthesis for children with an above-knee amputation. Gait data were collected for four subjects with above-knee prostheses walking at preferred and fast speeds. Joint moments were computed using anthropometrics from cadaver studies and direct measurements for each subject. The difference between these two methods primarily affected the inertia couple (Ialpha term) and the inertial effect due to gravity, which comprised a greater percentage of the total joint moment during swing as compared to stance. Peak hip and knee flexor and extensor moments during swing were significantly greater when calculated using cadaver data (p<0.05). These differences were greater while walking fast as compared to slow speeds. A significant difference was not found between these two methods for peak hip and knee moments during stance. A significant difference was found for peak ankle joint moments during stance, but the magnitude was not clinically important. These results support the use of direct measurements of anthropometry when examining above-knee prosthetic gait, particularly during swing.     

Two-dimensional video-based analysis of human gait using pose estimation

Human gait analysis is often conducted in clinical and basic research, but many common approaches (e.g., three-dimensional motion capture, wearables) are expensive, immobile, data-limited, and require expertise. Recent advances in video-based pose estimation suggest potential for gait analysis using two-dimensional video collected from readily accessible devices (e.g., smartphones). To date, several studies have extracted features of human gait using markerless pose estimation. However, we currently lack evaluation of video-based approaches using a dataset of human gait for a wide range of gait parameters on a stride-by-stride basis and a workflow for performing gait analysis from video. Here, we compared spatiotemporal and sagittal kinematic gait parameters measured with OpenPose (open-source video-based human pose estimation) against simultaneously recorded three-dimensional motion capture from overground walking of healthy adults. When assessing all individual steps in the walking bouts, we observed mean absolute errors between motion capture and OpenPose of 0.02 s for temporal gait parameters (i.e., step time, stance time, swing time and double support time) and 0.049 m for step lengths. Accuracy improved when spatiotemporal gait parameters were calculated as individual participant mean values: mean absolute error was 0.01 s for temporal gait parameters and 0.018 m for step lengths. The greatest difference in gait speed between motion capture and OpenPose was less than 0.10 m s⁻¹. Mean absolute error of sagittal plane hip, knee and ankle angles between motion capture and OpenPose were 4.0°, 5.6° and 7.4°. Our analysis workflow is freely available, involves minimal user input, and does not require prior gait analysis expertise. Finally, we offer suggestions and considerations for future applications of pose estimation for human gait analysis.