Sparse classification of discriminant nystagmus features using combined video-oculography tests and pupil tracking for common vestibular disorder recognition. Issue 4 (29th June 2020)

• Record Type: Journal Article
• Authors: Mouelhi, Aymen; Ben Slama, Amine; Marrakchi, Jihene; Trabelsi, Hedi; Sayadi, Mounir; Labidi, Salam
• Is Part Of: Computer methods in biomechanics and biomedical engineering. Volume 24:Issue 4(2021)
• Page Start: 400
• Page End: 418
• Publication Date: 2020-06-29
• Subjects: Vertigo -- nystagmus -- video-oculography tests -- vestibular disorders -- linear discriminant analysis -- sparse representation; Biomechanics -- Data processing -- Periodicals; Biomedical engineering -- Periodicals; Biomechanics -- Periodicals; Biomedical Engineering -- methods -- Periodicals; Computing Methodologies -- Periodicals; 612.7
• DOI: 10.1080/10255842.2020.1830972
• Languages: English
• ISSNs: 1025-5842

Abstract

Abstract: Vertigo is a common sign related to a problem with the brain or vestibular system. Detection of ocular nystagmus can be a support indicator to distinguish different vestibular disorders. In order to get reliable and accurate real time measurements from nystagmus response, video-oculography (VOG) plays an important role in the daily clinical examination. However, vestibular diseases present a large diversity in their characteristics that leads to many complications for usual analysis. In this paper, we propose a novel automated approach to achieve both selection and classification of nystagmus parameters using four tests and a pupil tracking procedure in order to give reliable evaluation and standardized indicators of frequent vestibular dysfunction that will assist clinicians in their diagnoses. Indeed, traditional tests (head impulse, caloric, kinetic and saccadic tests) are applied to obtain clinical parameters that highlight the type of vertigo (peripheral or central vertigo). Then, a pupil tracking method is used to extract temporal and frequency nystagmus features in caloric and kinetic sequences. Finally, all extracted features from the tests are reduced according to their high characterization degree by linear discriminant analysis, and classified into three vestibular disorders and normal cases using sparse representation. The proposed methodology is tested on a database containing 90 vertiginous subjects affected by vestibular Neuritis, Meniere's disease and Migraines. The presented technique highly reduces labor-intensive workloads of clinicians by producing the discriminant features for each vestibular disease which will significantly speed up the vertigo diagnosis and provides possibility for fully computerized vestibular disorder evaluation.