FACED-UP: Fast Analysis of CErvical Dystonia * Unobtrusive and Precise

Cervical dystonia (CD) is a chronic neurological condition that is
characterized by intermittent or continuous muscle contractions leading to
abnormal, often repetitive, movements and/or postures. This is often associated
with pain and substantial limitations in (both physical and social) daily life
functioning. For diagnosis and treatment selection, CD is mainly evaluated
using clinical scales and questionnaires, such as the Toronto Western Spasmodic
Torticollis Rating Scale (TWSTRS). These evaluations are subjective and
strongly depend on the expertise of the clinician. Sometimes CD is misdiagnosed
as essential tremor (ET), as a consequence of which patients may incorrectly be
injected with botulinum toxin (BTX) or, vice versa, unnecessarily be exposed to
ineffective medication. Hence, there is a need for objective and user-friendly
techniques that accurately capture the characteristics of CD. Emerging
technologies such as unobtrusive motion tracking and computer vision combined
with pattern recognition techniques offer the required breakthrough
opportunities to address the shortcomings of traditional evaluations of CD.

The primary objective of this study is to develop an objective,
patient-friendly assessment setup with which relevant characteristics of
movement disorders of the head, such as CD, can be obtained from video- and
depth-images (obtained using Microsoft KinectTM) without markers and/or sensors
being attached to the patients. The specific objectives are: 1) to develop
computer-vision techniques for accurate evaluation of postural deviations and
movement disorders (trembling, shaking) of the head from video- and depth
images obtained with KinectTM cameras, and to establish whether this markerless
motion capture system is able to capture the characteristic features of CD,
i.e., whether differences between CD patients and control subjects can be
detected for the concerning parameters; 2) to establish the association between
outcome parameters derived from the markerless motion tracking system and the
corresponding (sub)scores on clinical tests (i.e., the TWSTRS score) for CD
patients; 3) to establish whether the markerless motion capture system is
sensitive enough for 3D-kinematic measurement of the head (compared to
reference motion capture systems using accelerometers and position-orientation
sensors); and 4) to explore whether machine learning and pattern recognition
techniques can be used to distinguish CD patients from ET patients.

Observational, cross-sectional study in which parameters of postures and
movements of the head are collected in two groups of patients (CD and ET) and
healthy controls.

In the Technology in Motion Laboratory at the LUMC, a standard video assessment
protocol will be conducted according to the TWSTRS guidelines.The protocol
starts with a recording at rest, followed by several head movements to
determine the maximum active range of motion and recording of a *sensory
trick* (if present). The participant*s movements are recorded with a markerless
motion capture device (i.e., Microsoft KinectTM sensor) complemented with two
small accelerometers and two position-orientation sensors for validation
purposes. Based on these video recordings, the TWSTRS severity scale will be
completed by an experienced neurologist. Computer vision and
pattern-recognition techniques will be applied to the high-resolution color-
and depth data obtained from the Microsoft KinectTM to accurately capture
characteristic features of CD.

The FACED-UP protocol will align with current clinical practice (i.e., will not
interfere with clinical decision making). FACED-UP will add a short
questionnaire (expected duration 5-10 minutes) and a measurement (expected
duration 20 minutes) using a markerless tracking device for assessment of
postures and movements of the head. When possible, measurements are performed
during regular visits to the outpatient clinic, such that no extra visits of
the patients are required. Measurements are non-invasive and bear minimal
risks. Active participation of the subjects is required, but tasks will mostly
involve submaximal effort. Travel costs to the hospital will be compensated on
the basis of public transport (2nd class) or travel distance (¤ 0.19 per km).
Additionally, participants will receive a VVV-voucher to the value of ¤20 for
their participation.