Quantitative Evaluation of Cerebellar Ataxia through Automated Assessment of Upper Limb Movements

Cerebellar damage can result in peripheral dysfunction manifesting as poor and inaccurate coordination, irregular movements, and tremors. Conventionally, the severity assessment of Cerebellar ataxia (CA) is primarily based on expert clinical opinion and hence likely to be subjective. In order to establish inter-rater concordance with enhanced reliability and effectiveness in the assessment of upper limb function, a novel automated system employing Microsoft Kinect © is considered to capture the motion of the patient’s finger for objective assessment. This essentially mimics the commonly used finger tracking task clinically assessed through subjective observation. A clinical trial was conducted involving 42 CA patients and 18 age-matched healthy subjects. The relevant kinematically diagnostic features of CA patients allowed a classification accuracy of 97% using the Bayesian Quadratic Discriminant Analysis (QDA). The correlation (severity) between the extracted features and the independent severity scores from expert clinicians were collated to achieve a high correlation ( ${r} = {0.86}$ , ${p} ) with the Scale for the Assessment and Rating of Ataxia (SARA). The proposed system can efficiently generate objective information of severity as a result of features that are not necessarily observable during standard bedside clinical testing. Furthermore, the superior performance of the Ballistic (finger chase) test indeed supports the credence of the Ramp test redundancy that exists among the wider clinical community.