Gait Motion Classification for Neurodegenerative Diseases by Recurrence Structure Analysis - Ecole Nationale du Génie de l'Eau et de l'Environnement de Strasbourg
Pré-Publication, Document De Travail (Preprint/Prepublication) Année : 2024

Gait Motion Classification for Neurodegenerative Diseases by Recurrence Structure Analysis

Résumé

Objective: Gait analysis plays a significant role in clinical assessments to discriminate neurological disorders from healthy controls, to grade disease severity, and to further differentiate dementia subtypes. Methods: In this paper, we propose to apply recurrence structure analysis (RSA) as a method to classify pathological gait tasks from 3D skeleton pose sequences. Results: For each dataset, RSA yields symbolic sequences, whose complexity reflects the subject's gait movement complexity. A new gait movement model permitted to derive novel complexity measures, which serve as classification features. Applying a Multi-Layer Perceptron classification to healthy and pathological subjects suffering from Alzheimer Disease (AD) or Dementia with Lewy Bodies (DLB) permits to distinguish subjects with regular and irregular gait and AD and DLB patients. Moreover, the performed analysis indicates that arms movement is more informative in the distinction of AD and DLB than legs movement. A final comparison to previous studies of similar data demonstrates that the proposed model-based feature classification outperforms some previous data-based feature classification methods. Conclusion: Model-based data features permit to discriminate patients suffering from Alzheimer Disease and Dementia with Lewy Bodies based on gait videos. Significance: Complexity model features derived from gait videos permit the classification of neurological patients, which outperforms some data-based feature classification methods.
Fichier principal
Vignette du fichier
bare_jrnl_transmag.pdf (694.58 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04707101 , version 1 (24-09-2024)

Licence

Identifiants

  • HAL Id : hal-04707101 , version 1

Citer

Gauthier Debes, Diwei Wang, Peter beim Graben, Frédéric Blanc, Candice Muller, et al.. Gait Motion Classification for Neurodegenerative Diseases by Recurrence Structure Analysis. 2024. ⟨hal-04707101⟩
0 Consultations
0 Téléchargements

Partager

More