Objective: We compared markerless video-based motion analysis with inertial measurement unit (IMU)-based kinematic recordings in healthy controls and patients with CD.
Background: Cervical dystonia (CD) is characterised by involuntary contractions of neck muscles resulting in abnormal head postures. Diagnosis and characterisation of the primary muscles driving symptoms are based on expert clinical evaluation, often aided by video recordings. Quantitative analysis of video data may enable objective measurement of dystonic movements and assessment of treatment response.
Method: Head and neck movements were recorded simultaneously using three modalities: a Delsys IMU-based kinematic system, a 3D depth camera, and 2D standard video recording. Markerless pose estimation was performed offline using OpenPose and online using Nuitrack software. Kinematic outputs from the video-based systems were compared against the IMU. Displacements, velocities and derived kinematic parameters were analysed across modalities in both healthy controls and patients with CD.
Results: Cyclic head rotation, lateral flexion and isolated head rotation were chosen to assess cervical motion across three kinematic modalities in four healthy controls. Bland-Altman analysis demonstrated consistent superiority of markerless pose estimation over the 3D depth camera when compared to the IMU ground reference. Lateral flexion presented the closest approximation to ground truth for markerless pose estimation (mean difference = 3.250°, 95% Limit of Agreement (LoA): -23.35° to 29.85°), compared to the 3D depth camera (mean difference = 30.95°, 95% LoA: 24.33° to 37.57°).
Conclusion: Markerless video-based approaches provide a feasible method for quantifying head movements relevant to CD. While both depth camera and smartphone-based recordings showed systematic underestimation relative to IMU-derived measurements, these approaches may enable scalable, low-cost quantification of dystonic movements and facilitate objective monitoring of disease severity and treatment response.
To cite this abstract in AMA style:
O. Killian, S. Jimenez, A. Gill, C. Fearon, L. Williams, S. O'Riordan, M. Hutchinson, R. Reilly. A Comparison of Video-Based and Sensor-Based Movement Measurements in Cervical Dystonia [abstract]. Mov Disord. 2026; 41 (suppl 1). https://www.mdsabstracts.org/abstract/a-comparison-of-video-based-and-sensor-based-movement-measurements-in-cervical-dystonia/. Accessed October 1, 2026.« Back to 2026 International Congress
MDS Abstracts - https://www.mdsabstracts.org/abstract/a-comparison-of-video-based-and-sensor-based-movement-measurements-in-cervical-dystonia/