Objective: Identify patterns of changes in neck muscle configuration in cervical dystonia using automated analysis of ultrasound data.

Background: Ultrasound is a promising tool in the assessment and management of cervical dystonia. We have previously shown that automated identification of neck muscle boundaries using ultrasound is possible in healthy control participants [1], and developed a method for machine-learning based segmentation in people with cervical dystonia [2].

Method: We studied 35 patients with idiopathic cervical dystonia (mean age 61±10 years) and 26 healthy controls (mean age 59±14 years). All participants underwent real-time ultrasound imaging of the posterior neck muscles. Following automated segmentation of the neck muscles using machine learning, a principal components analysis of changes in muscle shape and configuration was performed.

Results: Principal components analysis of the segmented ultrasound data showed significant differences between participants with dystonia and healthy controls when comparing the first two eigenvalues (P<0.0001). Two distinct clusters of muscle shape in people with dystonia were evident on the analysis, compared to healthy controls.

Conclusion: Automated analysis of muscle shape from real-time ultrasound data provides useful diagnostic information which can differentiate cervical dystonia from healthy controls.

References: 1. Cunningham RJ, Harding PJ, Loram ID. Real-time ultrasound segmentation, analysis and visualization of deep cervical muscle structure. IEEE Trans Med Imaging 2017;36:653-665. 2. Siddique AB, Cunningham RJ, Silverdale MA, Harding PJ, Loram ID, Kobylecki C. Mov Disord 2018;33 suppl 2: 322.

« Back to 2019 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/automated-analysis-of-neck-muscle-ultrasound-can-distinguish-cervical-dystonia-from-healthy-controls/