Researcher: Wei Tang, MSc (PhD student)
Expected thesis defense: 2025
Funding: Chinese Scholarship Council
“Ataxia” refers to the impairment of the expected smooth performance of targeted directional movements, leading to impaired coordination, such as instability when walking, limbs shaking, slow response and poor accuracy. The recognition of ataxia with onset in early life (EOA) is difficult because children with Developmental Coordination Disorder (DCD) present with phenotypically impaired coordination that resembles ataxia, exhibiting slow motor behavior in the absence of mental retardation and visual impairment. Children with DCD often avoid social interaction and physical activity and suffer from many secondary psychosocial problems such as inferiority and depression. Thus, the diagnosis of DCD and EOA is very important for current and future management and health. Currently, clinicians usually use the Scale for the Assessment and Rating of Ataxia (SARA) to measure the severity of ataxia. But researchers found that phenotypic discrimination between EOA and DCD is incomplete. Therefore, the use of inertial measurement units, (IMUs) attached to the body with elastic straps including accelerometers and gyroscopes obtained during movements for automatic classification, has been studied for their use as an aid in the differential diagnosis of EOA and DCD. Mannini et al (2017) found that automatic classification based on quantitative gait features performs better than classification based on the phenotypic diagnosis, which suggests that quantification of movement and subsequent automatic classification could provide a support tool for consistent and repeatable diagnostic evaluation. Besides, Martinez- Manzanera et al. (2018) found that the variation within finger to nose trajectories was also different between EOA and DCD. Yet, the different studies into this matter have shown that DCD remains poorly recognizable, with overlap between ataxia and DCD coordination disorders, as well as between physiologically immature coordination and DCD, both phenotypically and when classifying on the basis of quantified movements using inertial measurement units (IMUs). In the study proposed here, the aim is to achieve further improvement in the differentiation between DCD and EOA in a clinical setting using video-based movement quantification with deep neural networks.