Article Dans Une Revue Pattern Analysis and Applications Année : 2025

Learning a geometric deep representation to classify Parkinson smooth pursuit patterns

Résumé

Parkinson's disease (PD) is characterized by the degeneration of dopaminergic neurotransmitters, leading to motor disturbances that typically become apparent in advanced stages of the disease. Currently, Smooth Pursuit Eye Movement (SPEM) analysis has been suggested to be a potential biomarker for PD. However, conventional recording SPEM methods involve intrusive procedures, specialized protocols, and mainly provide information based on a single global displacement trajectory. We hypothesize that SPEM patterns emcompass a diverge range of movement, characterized by intricate spatio-temporal relationships, which are crucial for quantifying Parkinsonian related patterns, even at early stages. This work introduces a novel end-to-end deep learning representation model that encodes spatio-temporal SPEM patterns and captures geometric second-order relationships to differentiate between PD and control subjects. The geometric learning scheme considers a Riemmanian manifold structure from the spatiotemporal deep activations resulting from 3D volumetric convolutions of a set of video recording. Following a non-intrusive video-based recording protocol, the proposed approach achieved an excellent AUC-ROC score across several SPEM tasks configurations, with a total of 22 subjects (11 control and 11 PD patients) participating in the study. The geometrical learning has the ability to encode relevant spatio-temporal SPEM relationships that strongly support the discrimination of Parkinsonim related patterns.

Fichier principal
Vignette du fichier
PAA_Fernando25.pdf (929.94 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-05430038 , version 1 (23-12-2025)

Licence

Identifiants

Citer

Luis Fernando Celis, Juan Olmos, Antoine Manzanera, Fabio Martínez. Learning a geometric deep representation to classify Parkinson smooth pursuit patterns. Pattern Analysis and Applications, 2025, 28 (3), pp.153. ⟨10.1007/s10044-025-01514-w⟩. ⟨hal-05430038⟩
60 Consultations
66 Téléchargements

Altmetric

Partager

  • More