Interpretable polar-flow decomposition for torsional nystagmus lateralization in posterior-canal BPPV
- 1School of Disaster and Emergency Medicine, Tianjin University , ,
- 2Affiliated Wenzhou People’s Hospital of Hangzhou Medical College, Wenzhou People’s Hospital, Third Clinical College of Wenzhou Medical University , , ,
Abstract
Objective: Torsional nystagmus during the Dix–Hallpike maneuver represents a cardinal vestibulo-ocular reflex (VOR) sign in posterior-canal benign paroxysmal positional vertigo (pc-BPPV). This study develops an interpretable framework for automated classification of torsional nystagmus direction and temporal localization of abnormal events from clinical infrared recordings.
Methods: Infrared Dix–Hallpike videos were processed via You Only Look Once version 8 nano (YOLOv8n) pupil localization, dynamic iris masking, and Recurrent All-Pairs Field Transforms (RAFT) optical flow. Flow vectors were projected into pupil-centered polar coordinates to extract tangential and radial motion features. A bidirectional gated recurrent unit with attention (BiGRU-attention) network performed segment-level three-class classification with sliding-window patient-level inference.
Results: Five-fold cross-validation yielded segment-level accuracy of 94.19% ± 0.88% with balanced F1-scores across all three classes. On an independent 45-patient long-video test set, patient-level accuracy reached 97.78% at 60 frames per second (fps) and 100% (45/45) at 30 fps. Temporal localization achieved mean intersection-over-union (IoU) of 0.899 and 0.896, with hit rates of 96.8 and 100%, respectively.
Conclusion: The polar-flow framework enables physiologically interpretable and robust classification of pc-BPPV-related torsional-upbeat nystagmus. By explicitly decomposing eye movements into tangential and radial components, it provides reliable lateralization while demonstrating the potential for clinically meaningful temporal localization of abnormal nystagmus events.
Related
- Beat-to-Beat Nystagmus Differentiation Between Benign Paroxysmal Positional Vertigo (BPPV) and Vestibular Migraine (VM): A Cross-Sectional Videonystagmographic Study From Northern India
- Vertigo and Nystagmus: The Role of the Velocity Storage Integrator in Skull Vibration Induced Nystagmus (SVIN).
- Case Report: A novel intronic FRMD7 variant associated with X-linked infantile nystagmus: functional validation by a minigene splicing assay
- Transient hemispheric lateralization is associated with intensity-dependent aftereffects of low-frequency rTMS
- Behavioral markers of motor reserve: stability, lateralization, and consistency
BibTeX
@article{li2026interpretable,
title = {{Interpretable polar-flow decomposition for torsional nystagmus lateralization in posterior-canal BPPV}},
author = {Yanjun Li and Xiaomin Xu and Lidan Gao and Qiqi Shao and Jie Chen and Wenhuan Wang and Xiaoqing Li and Congcong Sun and Shike Hou and Xiaokai Yang},
journal = {Frontiers in neurology},
year = {2026},
month = oct,
doi = {10.3389/fneur.2026.1963383},
url = {https://europepmc.org/article/PMC/PMC13671929},
note = {CC BY},
}
RIS
TY - JOUR
TI - Interpretable polar-flow decomposition for torsional nystagmus lateralization in posterior-canal BPPV
AU - Yanjun Li
AU - Xiaomin Xu
AU - Lidan Gao
AU - Qiqi Shao
AU - Jie Chen
AU - Wenhuan Wang
AU - Xiaoqing Li
AU - Congcong Sun
AU - Shike Hou
AU - Xiaokai Yang
PY - 2026
DA - 2026/10/02
JO - Frontiers in neurology
DO - 10.3389/fneur.2026.1963383
UR - https://europepmc.org/article/PMC/PMC13671929
AB - Objective: Torsional nystagmus during the Dix–Hallpike maneuver represents a cardinal vestibulo-ocular reflex (VOR) sign in posterior-canal benign paroxysmal positional vertigo (pc-BPPV). This study develops an interpretable framework for automated classification of torsional nystagmus direction and temporal localization of abnormal events from clinical infrared recordings. Methods: Infrared Dix–Hallpike videos were processed via You Only Look Once version 8 nano (YOLOv8n) pupil localization, dynamic iris masking, and Recurrent All-Pairs Field Transforms (RAFT) optical flow. Flow vectors were projected into pupil-centered polar coordinates to extract tangential and radial motion features. A bidirectional gated recurrent unit with attention (BiGRU-attention) network performed segment-level three-class classification with sliding-window patient-level inference. Results: Five-fold cross-validation yielded segment-level accuracy of 94.19% ± 0.88% with balanced F1-scores across all three classes. On an independent 45-patient long-video test set, patient-level accuracy reached 97.78% at 60 frames per second (fps) and 100% (45/45) at 30 fps. Temporal localization achieved mean intersection-over-union (IoU) of 0.899 and 0.896, with hit rates of 96.8 and 100%, respectively. Conclusion: The polar-flow framework enables physiologically interpretable and robust classification of pc-BPPV-related torsional-upbeat nystagmus. By explicitly decomposing eye movements into tangential and radial components, it provides reliable lateralization while demonstrating the potential for clinically meaningful temporal localization of abnormal nystagmus events.
SN - 1664-2295
ER -