Entropy-based analysis of rhythmic tapping for the quantitative assessment of cerebellar ataxia. (May 2020)

• Record Type: Journal Article
• Authors: Nguyen, Khoa D.; Pathirana, Pubudu N.; Horne, Malcolm; Power, Laura; Szmulewicz, David J.
• Is Part Of: Biomedical signal processing and control. Volume 59(2020)
• Publication Date: 2020-05
• Subjects: Wearable inertial system -- Mobile health informatics -- Cerebellar ataxia -- Rhythmic tapping -- Information theory; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2020.101916
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: A wearable inertial system was used to effectively estimate the severity of cerebellar ataxia in a simple exercise of rhythmic tapping. Disabilities in tapping were manifested in multiple domains (time, frequency, phase space). The complexity loss hypothesis was affirmed in rhythmic tapping and carried across multiple scaling dimensions. Spectral entropy can measure inadvertent movements in the frequency range dependent on the self-paced tapping rate. Abstract: The severity of appendicular ataxia is currently assessed by neurological specialists observing the motor deficits during the performance of bedside tests. These tests are traditionally restricted to discrete subjective scales and prescribed by constraints that may in turn affect the subject's performance. To overcome these limitations and facilitate uniform monitoring of cerebellar ataxia, an instrumented assessment system using a wearable sensor has been developed for the rhythmic finger and foot tapping tasks. Departing from the conventional obligation to spatiotemporal parameters that are described clinically, the proposed approach investigates features based on information theory to evaluate the variability of the tapping dynamics. From a clinical trial of 24 healthy subjects and 34 patients diagnosed with cerebellar ataxia (CA), the findings revealed not only are there more variations in movement patterns and timing of tapping events in CA but that movement complexity was also degraded across multiple entropy scales. The system using combined models of different features was able to achieve an average classification accuracy of 81% and 79% in finger and foot tapping respectively. The outcomes strongly agreed with the clinical scale widely-used for the assessment and rating of ataxia (SARA) to a correlation level of 0.71 ( p < 0.001) in finger tapping and 0.59 ( p < 0.001) in foot tapping.