Vision-based Assessment of Joint Extensibility in Ehlers Danlos Syndrome

Assessing the Feasibility of a Smartphone-based, Machine Learning Visual Imaging Application for Assessment of Hyperextensibility of Peripheral Joints in Ehlers Danlos Syndrome

Ehlers Danlos Syndrome (EDS) is a heterogenous group of genetic disorders with 13 identified subtypes. Hypermobile EDS (hEDS), although the most common subtype of EDS, does not yet have an identified genetic mutation for diagnostic confirmation. Generalized joint hypermobility (GJH) is one of the hallmark features of hEDS. The scoring system used in measurement of GJH was described by Beighton. The Beighton score is calculated using a dichotomous scoring system to assess the extensibility of nine joints. Each joint is scored as either hypermobile (score = 1) or not hypermobile (score = 0). The total score (Beighton score) can vary between a minimum of 0 and a maximum of 9, with higher scores indicating greater joint laxity.

While there is moderate validity and inter-rater variability in using the Beighton score, there continue to be several challenges with its widespread and consistent application by clinicians. Some of the barriers reported in the literature include:

i) In open, non-standardized systems there can be significant variation in the method to perform these joint extensibility tests including assessing baseline measurements, ii) Determining consistent and standard measurement tools/methodology e.g. goniometer use can vary widely iii) Assessing the reliability of the cut off values and, iv) Performing full assessment prior to informing patients of possible classification of GJH positivity (low specificity and low positive predictive).

Inappropriate implementation of tests to assess GJH results in inaccurate identification of GJH and potentially unintended negative consequences of making the wrong diagnosis of EDS. The objective of this study is to create a more robust and valid method of joint mobility measurement and reduce error in the screening of EDS through use of a smartphone-based machine learning application systems for measurement of joint extensibility.

The project will:

i) Create a smart-phone enabled visual imaging app to assess the measurement of joint extensibility, ii) Assess the feasibility of using the smart-phone app in a clinical setting to screen potential EDS patients, iii) Determine the validity of the application in comparison to in person clinical assessment in a tertiary care academic EDS program. If successful, the smart-phone application could help standardize the care of potential EDS patients in an efficient and cost-effective manner.

Study Overview

• Status: Enrolling by invitation
• Conditions: Ehlers-Danlos Syndrome
• Intervention / Treatment: Other: No intervention, additional video data collection only

• Study Type: Observational
• Enrollment (Estimated): 225

Contacts and Locations

Study Locations

• Canada
  • Ontario
    • Toronto, Ontario, Canada, M5G 2C4
      • GoodHope EDS - Toronto General Hospital

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: 18 years and older (Adult, Older Adult)
• Accepts Healthy Volunteers: No

• Sampling Method: Non-Probability Sample

Study Population

The population being studied includes all patients referred to or seen in the GoodHope EDS clinic. The clinic accepts referrals from symptomatic adult patients (age > 18 years), with EDS, or suspected EDS. EDS is a connective tissue disorder with 100% penetrance, but variable in phenotypic expression, suspected cases of EDS or G-HSD may therefore include other hereditary or acquired connective tissue diseases/disorder, and/or complex chronic illnesses characterized by, or that feature, joint hypermobility, pain, and fatigue.

Description

Inclusion Criteria:

• All patients seen in the GoodHope EDS clinic at Toronto General are eligible for inclusion, regardless of their presenting diagnosis or the results of their assessments

Exclusion Criteria:

• Patients who do not consent to participate will not be included (participants may withdraw consent at any time)

Study Plan

How is the study designed?

Design Details

• Observational Models: Case-Only
• Time Perspectives: Cross-Sectional

Number of groups / cohorts

1

Cohorts and Interventions

Group / Cohort	Intervention / Treatment
New patients at the GoodHope EDS clinic at Toronto General Hospital; All patients seen in the EDS clinic are eligible for inclusion, regardless of their presenting diagnosis or the results of their assessments.	Other: No intervention, additional video data collection only; No intervention will be used. Consenting participants will have video recordings taken during their exam of joint hypermobility which will be analyzed at a later time

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Comparison of agreement in predicted angle by pose-estimation library	The performance of the developed machine learning models for predicting the range of motion will be analyzed by the pose-estimation library used. This analysis will be performed on the subset of the data collected during the first 2 months of data collection. This information will be used to select the pose-estimation libraries to proceed with when refining the machine learning models.	4 months
Comparison of agreement in predicted angle by joint	The performance of the developed machine learning models for predicting the range of motion at each joint (spine, knee, ankle, elbow, shoulder, thumb, fifth finger) will be analyzed independently for each joint. This will provide insight with respect to which joints the system is more accurate at predicting from video.	1 year
Assess the accuracy of range of motion prediction using vision-based data	Machine learning models trained on videos of individuals performing the joint hypermobility maneuvers will be developed. Their performance will be compared to the range of motion measured by an expert clinician using a goniometer.	1 year

Collaborators and Investigators

• Sponsor: University Health Network, Toronto
• Investigators: Principal Investigator: Nimish Mittal, MD, GoodHope Ehlers Danlos Syndrome Clinic, Toronto General Hospital

Publications and helpful links

General Publications

• Critical Care Services Ontario, Ehlers-Danlos Syndrome Expert Panel Report, 2016. https://www.health.gov.on.ca/en/common/ministry/publications/reports/eds/Default.aspx.
• Cahill SV, Sharkey MS, Carter CW. Clinical assessment of generalized ligamentous laxity using a single test: is thumb-to-forearm apposition enough? J Pediatr Orthop B. 2021 May 1;30(3):296-300. doi: 10.1097/BPB.0000000000000732.
• He K, Gkioxari G, Dollár P, Girshick R. Mask R-CNN. In: Proceedings of the IEEE international conference on computer vision. 2017. p. 2961-9.
• Cao Z, Hidalgo G, Simon T, Wei SE, Sheikh Y. OpenPose: Realtime Multi-Person 2D Pose Estimation Using Part Affinity Fields. IEEE Trans Pattern Anal Mach Intell. 2021 Jan;43(1):172-186. doi: 10.1109/TPAMI.2019.2929257. Epub 2020 Dec 4.
• Fang H-S, Xie S, Tai Y-W, Lu C. RMPE: Regional Multi-person Pose Estimation. 2016 Nov 30; Available from: http://arxiv.org/abs/1612.00137
• Lin T-Y, Maire M, Belongie S, Bourdev L, Girshick R, Hays J, et al. Microsoft COCO: Common Objects in Context. 2014 May 1; Available from: http://arxiv.org/abs/1405.0312
• Andriluka M, Pishchulin L, Gehler P, Schiele B. 2D Human Pose Estimation: New Benchmark and State of the Art Analysis. In: 2014 IEEE Conference on Computer Vision and Pattern Recognition. 2014. p. 3686-93
• Lugaresi C, Tang J, Nash H, McClanahan C, Uboweja E, Hays M, et al. MediaPipe: A Framework for Building Perception Pipelines. 2019 Jun 14; Available from: https://arxiv.org/abs/1906.08172
• Zhang F, Bazarevsky V, Vakunov A, Tkachenka A, Sung G, Chang C-L, et al. MediaPipe Hands: On-device Real-time Hand Tracking. 2020 Jun 17; Available from: http://arxiv.org/abs/2006.10214
• Mehdizadeh S, Nabavi H, Sabo A, Arora T, Iaboni A, Taati B. Concurrent validity of human pose tracking in video for measuring gait parameters in older adults: a preliminary analysis with multiple trackers, viewing angles, and walking directions. J Neuroeng Rehabil. 2021 Sep 15;18(1):139. doi: 10.1186/s12984-021-00933-0.
• Sabo A, Mehdizadeh S, Ng KD, Iaboni A, Taati B. Assessment of Parkinsonian gait in older adults with dementia via human pose tracking in video data. J Neuroeng Rehabil. 2020 Jul 14;17(1):97. doi: 10.1186/s12984-020-00728-9.
• Lu M, Zhao Q, Poston KL, Sullivan EV, Pfefferbaum A, Shahid M, Katz M, Kouhsari LM, Schulman K, Milstein A, Niebles JC, Henderson VW, Fei-Fei L, Pohl KM, Adeli E. Quantifying Parkinson's disease motor severity under uncertainty using MDS-UPDRS videos. Med Image Anal. 2021 Oct;73:102179. doi: 10.1016/j.media.2021.102179. Epub 2021 Jul 21.
• Williams S, Zhao Z, Hafeez A, Wong DC, Relton SD, Fang H, Alty JE. The discerning eye of computer vision: Can it measure Parkinson's finger tap bradykinesia? J Neurol Sci. 2020 Sep 15;416:117003. doi: 10.1016/j.jns.2020.117003. Epub 2020 Jun 30.
• Ota M, Tateuchi H, Hashiguchi T, Kato T, Ogino Y, Yamagata M, Ichihashi N. Verification of reliability and validity of motion analysis systems during bilateral squat using human pose tracking algorithm. Gait Posture. 2020 Jul;80:62-67. doi: 10.1016/j.gaitpost.2020.05.027. Epub 2020 May 25.
• Slembrouck M, Luong H, Gerlo J, Schütte K, Van Cauwelaert D, De Clercq D, et al. Multiview 3d Markerless Human Pose Estimation from Openpose Skeletons. In: International Conference on Advanced Concepts for Intelligent Vision Systems. Springer; 2020. p. 166-78.
• Wang H, Xie Z, Lu L, Li L, Xu X. A computer-vision method to estimate joint angles and L5/S1 moments during lifting tasks through a single camera. J Biomech. 2021 Dec 2;129:110860. doi: 10.1016/j.jbiomech.2021.110860. Epub 2021 Nov 8.
• Yahya M, Shah JA, Warsi A, Kadir K, Khan S, Izani M. Real time elbow angle estimation using single RGB camera. 2018 Aug 21; Available from: https://arxiv.org/abs/1808.07017
• Shi B, Brentari D, Shakhnarovich G, Livescu K. Fingerspelling Detection in American Sign Language. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2021. p. 4166-75
• Kim I-H, Jung I-H. A Study on Korea Sign Language Motion Recognition Using OpenPose Based on Deep Learning. 디지털콘텐츠학회논문지 (Journal of Digital Contents Society). 2021;22(4):681-7.
• Levy HP. Hypermobile Ehlers-Danlos Syndrome. 2004 Oct 22 [updated 2018 Jun 21]. In: Adam MP, Feldman J, Mirzaa GM, Pagon RA, Wallace SE, Bean LJH, Gripp KW, Amemiya A, editors. GeneReviews(R) [Internet]. Seattle (WA): University of Washington, Seattle; 1993-2023. Available from http://www.ncbi.nlm.nih.gov/books/NBK1279/

Study record dates

Study Major Dates

• Study Start (Actual): April 26, 2022
• Primary Completion (Estimated): June 1, 2024
• Study Completion (Estimated): December 1, 2024

Study Registration Dates

• First Submitted: May 4, 2022
• First Submitted That Met QC Criteria: May 4, 2022
• First Posted (Actual): May 9, 2022

Study Record Updates

• Last Update Posted (Actual): November 28, 2023
• Last Update Submitted That Met QC Criteria: November 27, 2023
• Last Verified: November 1, 2023

More Information

Terms related to this study

• Keywords: Smartphone; Video; Joint Range of Motion; Ehlers-Danlos Syndrome; Computer-vision; Beighton Score
• Additional Relevant MeSH Terms: Pathologic Processes; Cardiovascular Diseases; Vascular Diseases; Skin Diseases; Disease; Congenital Abnormalities; Hematologic Diseases; Hemorrhagic Disorders; Genetic Diseases, Inborn; Connective Tissue Diseases; Hemostatic Disorders; Skin Diseases, Genetic; Skin Abnormalities; Collagen Diseases; Syndrome; Ehlers-Danlos Syndrome
• Other Study ID Numbers: 22-5073

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: NO
• IPD Plan Description: No IPD will be shared with other researchers.

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No