The measurement of in vivo joint angles during a squat using a single camera markerless motion capture system as compared to a marker based system
Introduction
Altered hip and knee mechanics have been attributed to numerous conditions across the lifespan. For example, hip and knee mechanics are routinely measured in children with cerebral palsy [e.g. [1]], adults who have their anterior cruciate ligament reconstructed [e.g. [2]], adults with patellofemoral pain [e.g. [3]], and older individuals with osteoarthritis [e.g. [4]]. Defining the alterations in hip and knee mechanics has been enhanced by the development of sophisticated marker-based motion capture systems. The ability of the clinician to have a reliable and accurate means of quantifying the same types of altered mechanics that researchers use is still lacking. This limits the translation of laboratory based findings of abnormal mechanics to the diagnosis and treatment of lower extremity injuries in the clinic. However, recent developments in markerless motion capture technology may offer the potential for such systems to be available to clinicians in the near term.
Marker-based systems have limitations that have driven the development of markerless motion capture technology. The main limitations being the use of many cameras that make it impractical to use in a variety of settings such as a patient's home, a clinic, or a sports field as well as the high cost of marker-based cameras. To address these limitations, single camera markerless systems have been evaluated and shown promise in their ability to measure finger kinematics [5], trunk lean [6], [7], and foot posture [8]. However, the measurement of hip and knee kinematics during movement is lacking and necessary to establish the Kinect as a clinical tool to aid in diagnosis and treatment. Therefore, the goal of this study was to compare the ability of a single camera markerless motion capture system to measure hip and knee angles during movement to those measured by a marker-based system.
Section snippets
Data collection
After completing an informed consent approved by an institutional review board, 15 healthy people participated in this study (8 male, 7 female, height 1.702 ± 0.089 m, mass 67.9 ± 10.4 kg, age 24 ± 4 years, BMI 23.4 ± 2.2 kg/m2). First, 28 retro-reflective markers were placed on the participant (Fig. 1). Next, the participant was scanned using a projector-camera setup [9], [10], which creates a three-dimensional surface model of the person. Then, marker trajectories were measured with a 10 camera motion
Results
The patterns of motion were similar between systems where the difference between systems was greatest at the peak flexed position (i.e. bottom of the squat) (Fig. 3). The Kinect underestimated peak hip flexion by 4.3 degrees (9% of the total range of hip flexion motion) (Table 1) where the bias of hip flexion angles was −6.5 deg (Table 2). No significant differences were found in peak hip adduction, axial rotation, or knee angles (Table 1), in which the bias was <7 deg (Table 2). Peak joint
Discussion
The use of a single Microsoft Kinect camera holds potential as a clinical surrogate for the assessment of 3D motion. In this study, we compared the ability of a Kinect to measure joint angles with current state of the art, a marker-based motion capture system. Both systems agreed well for the shape of the motion calculated (Fig. 3); had high between trial reliability (Table 1); and exhibited a strong correlation between systems for the peak angles calculated (Fig. 4).
The average absolute
Acknowledgement
This work was funded by the Division of Information and Intelligent Systems of the National Science Foundation, grant 1231545.
Conflict of interest statement: There are no conflicts of interest to disclose.
References (26)
- et al.
Proximal and distal kinematics in female runners with patellofemoral pain
Clin Biomech
(2012) - et al.
Association between in vivo knee kinematics during gait and the severity of knee osteoarthritis
Knee
(2012) - et al.
Validity of the Microsoft Kinect for providing lateral trunk lean feedback during gait retraining
Gait Posture
(2013) - et al.
Validity of the Microsoft Kinect for assessment of postural control
Gait Posture
(2012) - et al.
Preferred placement of the feet during quiet stance: development of a standardized foot placement for balance testing
Clin Biomech
(1997) - et al.
Reliability and minimal detectible change values for gait kinematics and kinetics in healthy adults
Gait Posture
(2012) - et al.
Statistical methods for assessing agreement between two methods of clinical measurement
Int J Nurs Stud
(2010) - et al.
Accuracy of the Microsoft Kinect sensor for measuring movement in people with Parkinson's disease
Gait Posture
(2014) - et al.
Analysis of knee joint kinematics during walking in patients with cerebral palsy through human motion capture and gait model-based measurement
- et al.
Abnormal rotational knee motion during running after anterior cruciate ligament reconstruction
Am J Sports Med
(2004)
Markerless motion capture and measurement of hand kinematics: validation and application to home-based upper limb rehabilitation
IEEE Trans Biomed Eng
Reliability and validity of the Microsoft Kinect for evaluating static foot posture
J Foot Ankle Res
Build your own 3D scanner: optical triangulation for beginners
Cited by (70)
The role of individual physical body measurements and activity on spine kinematics during flexion, lateral bending and twist tasks in healthy young adults – Comparing marker(less) data
2023, Biomedical Signal Processing and ControlCitation Excerpt :Sixty-two healthy adults (male and female) with no spinal injury or lower back pain or no clinical spine deformity were recruited. We applied vector analysis to compute angular displacement from marker-defined vector angles [63] as it can be applied to marker-based and marker-less systems [52,8,1] – abstract only). Considering the findings of [72,29], an s-shaped spinal curvature was assumed when extracting the angles.
Concurrent validity of evaluating knee kinematics using Kinect system during rehabilitation exercise
2021, Medicine in Novel Technology and DevicesCitation Excerpt :They need to take factors like maximum deviation from good posture or alignment during exercise into concern and encourage patients to diminish abnormal errors. The peak angle is also a commonly reported outcome measure in other studies [22,30]. Therefore, the average peak knee angle in the sagittal and coronal planes for each task were utilized for statistical analysis.
A new method for assessing landing kinematics in non-laboratory settings
2021, Physical Therapy in SportCitation Excerpt :Along with the automated video analysis methods used in the present study, there are markerless motion analysis methods available (e.g. Physimax, Northbrook, Illinois) that were designed for clinical settings and rely on one Microsoft Kinect camera. While some have found that Kinect cameras are accurate (Schmitz et al., 2015), other have found poor agreement between Kinect cameras and traditional marker-based 3D motion capture (Bonnechère et al., 2014; Mentiplay et al., 2015; Tamura et al., 2020). Future work is needed to directly compare the accuracy and feasibility of different single-camera motion analysis methods that are intended for clinical use.
Marker-less versus marker-based driven musculoskeletal models of the spine during static load-handling activities
2020, Journal of BiomechanicsCitation Excerpt :This low-cost, portable, marker-less system allows for easy posture measurements in real workplaces with no need to wear skin mounted markers/sensors that may interfere with the worker’s function. While a number of studies have investigated the accuracy of Kinect versus the traditional optical marker-based motion capture systems especially during gain trials (e.g., Bonnechère et al., 2014; Clark et al., 2012, 2013; Dutta, 2012; Eltoukhy et al., 2017a; Galna et al., 2014; Mentiplay et al., 2015; Mobini et al., 2014; Pfister et al., 2014; Reither et al., 2018; Schmitz et al., 2015; Xu et al., 2015), there are only few studies that have used Kinect in real workplace conditions using qualitative risk assessment tools such as OWAS (Diego-Mas and Alcaide-Marzal, 2014), RULA (Plantard et al., 2017a), and NIOSH (Patrizi et al., 2016). Kinect has also been used to drive musculoskeletal or inverse dynamics models but only for gait trials (Andersen et al., 2013; Eltoukhy et al., 2017b; Skals et al., 2017), shoulder abduction (Skals et al., 2017), symmetric squatting (Jun et al., 2003), and shoulder/elbow movements during getting/putting tasks (Plantard et al., 2017b).
Automated classification of movement quality using the Microsoft Kinect V2 sensor
2020, Computers in Biology and MedicineCitation Excerpt :The intelligent system used joint kinematic features calculated from Kinect V2 position data. Although the validity of Kinect position data had been highly debated [27–31], its reliability and consistency are indisputable [28,30,32–34]. Kinect V2 is a reliable instrument for collecting position data, and the position data led to the detection of kinematic differences between those with poor movement quality against those with good movement quality.