Hold that pose: capturing cervical dystonia's head deviation severity from video

Abstract

Objective Deviated head posture is a defining characteristic of cervical dystonia (CD). Head posture severity is typically quantified with clinical rating scales such as the Toronto Western Spasmodic Torticollis Rating Scale (TWSTRS). Because clinical rating scales are inherently subjective, they are susceptible to variability that reduces their sensitivity as outcome measures. The variability could be circumvented with methods to measure CD head posture objectively. However, previously used objective methods require specialized equipment and have been limited to studies with a small number of cases. The objective of this study was to evaluate a novel software system—the Computational Motor Objective Rater (CMOR)—to quantify multi‐axis directionality and severity of head posture in CD using only conventional video camera recordings. Methods CMOR is based on computer vision and machine learning technology that captures 3D head angle from video. We used CMOR to quantify the axial patterns and severity of predominant head posture in a retrospective, cross‐sectional study of 185 patients with isolated CD recruited from 10 sites in the Dystonia Coalition. Results The predominant head posture involved more than one axis in 80.5% of patients and all three axes in 44.4%. CMOR's metrics for head posture severity correlated with severity ratings from movement disorders neurologists using both the TWSTRS‐2 and an adapted version of the Global Dystonia Rating Scale (rho = 0.59–0.68, all p <0.001). Conclusions CMOR's convergent validity with clinical rating scales and reliance upon only conventional video recordings supports its future potential for large scale multisite clinical trials.

Quantitative assessment of head movement dynamics in dystonia using visual perceptive deep learning: a multi-centre retrospective longitudinal cohort study

“…We first demonstrate the robustness and clinical applicability of our deep learning framework in accurately inferring head-angle deviations during attempted ‘null’ head positions from diverse clinical videos captured using consumer-grade hardware. Our visual perceptive approach surpasses various vision-based frameworks that relied on multiple or specialized depth cameras to automate ratings [21, 22], and achieves comparable performance to a recent study by Zhang et al [20]. However, a distinguishing feature of our approach is the ability to estimate head angles in real-time using a portable device such as smartphones or tablets.…”

“…Whilst adaptions to TWSTRS have incorporated tremor related features [40, 41], there is still a growing demand for more objective and granular disease metrics [3, 18]. Already widely used for quantitative phenotyping in experimental neuroscience, computer vision approaches have recently emerged as a promising new tool for clinical assessments [20, 23, 24, 42].…”

“…The generalised dystonia cohort, used for independent validation, comprised 30 patients from the same centres. Individual datasets were included if (i) they contained at least one pre-operative clinical rating video showing the full dystonic phenotype and a video from the chronic postoperative phase (3-15 months post surgery) documenting the effects of clinically programmed DBS and (ii) both videos fulfilled minimal criteria ensuring video quality, which were chosen to reflect the current best practice in clinical computer vision approaches [20, 23, 24]. These were: (i) front view perspective of a single individual sitting on a chair, (ii) no significantly obscuring items on patients (e. g. excessive head dressings with externalised DBS device), (ii) no excessive camera movements, variable zoom depths or lighting insufficient to identify typical body landmarks (e.g., eyes), (iii) continuous presence of head and neck in the camera frame.…”

The use of digital outcome measures in clinical trials in rare neurological diseases: a systematic literature review

“…[ 49 ] Individuals with HD spent over 50% of the total time lying down, more than individuals with prodromal HD, PD, and controls [ 47 ] Cambridge Neurotechnology AW4 or Respironics Actiwatch 2 actigraph MPS Children with MPS III had significantly higher activity levels during the early morning hours compared to controls [ 117 ] CHDR Monitoring Remotely (CHDR MORE) platform FSHD The classification between patients with FSHD and controls with 93% accuracy, 100% sensitivity, and 80% specificity. Features relating to smartphone acceleration, app use, location, physical activity, sleep, and call behavior were the most salient features for the classification [ 39 ] Computational Motor Objective Rater (CMOR) FD Head posture severity correlated with severity ratings from movement disorders neurologists using both the TWSTRS-2 and an adapted version of the Global Dystonia Rating Scale [ 85 ] DynaPort Move Monitor, McRoberts, The Hague, The Netherlands MG Patients perform less vigorous PA, spend more time sedentary and engage in less and shorter durations of MVPA than controls. Habitual PA correlated positively with 6 min walking distance [ 41 ] PAL was lower in patients than in controls.…”

Improving the Efficacy of Botulinum Toxin for Cervical Dystonia: A Scoping Review

“…Moreover, they do not capture these novel aspects of the disorder including the COL-CAP concept [ 101 ]. This opens the question of how to measure the efficacy of BoNT with initial suggestions advocating the use of objective markers, at least to measure changes in the motor component of the disorder [ 102 ]. At the other end of the spectrum, however, one should consider that non-motor symptoms such as pain and mood dysfunction are integral to CD and might significantly influence the subjective perception of treatment efficacy [ 16 ].…”