Classification of dysphagia severity after lateral medullary infarction with deep learning.
Authors
Affiliations (4)
Affiliations (4)
- Department of Physical Medicine and Rehabilitation, Dongguk University Ilsan Hospital, College of Medicine, 27 Dongguk-ro, Ilsandong-gu, Goyang, 10326, Republic of Korea.
- Department of Otorhinolaryngology-Head and Neck Surgery, Dongguk University Ilsan Hospital, College of Medicine, 27 Dongguk-ro, Ilsandong-gu, Goyang, 10326, Republic of Korea.
- Graduate School of Management of Technology, Korea University, Seoul, 02841, Republic of Korea. [email protected].
- Department of Physical Medicine and Rehabilitation, Dongguk University Ilsan Hospital, College of Medicine, 27 Dongguk-ro, Ilsandong-gu, Goyang, 10326, Republic of Korea. [email protected].
Abstract
Dysphagia is a common and debilitating complication in patients with lateral medullary infarction (LMI), affecting up to 100% of cases and significantly impairing quality of life. Accurate classification of early dysphagia severity is essential for timely intervention and personalized rehabilitation planning. This study aimed to develop and validate a deep learning algorithm using acute-phase diffusion-weighted MRI to classify dysphagia severity in LMI patients. A retrospective cohort of 163 patients with confirmed acute LMI was analyzed. Dysphagia severity was determined by videofluoroscopic swallowing studies (VFSS), categorizing patients into severe and non-severe groups. Lesion regions were manually labeled and preprocessed for model training. Transformer-based deep learning architecture, the Hierarchical Vision Transformer (Hier-ViT), was employed due to its capacity to model spatial hierarchies and global image context. The model achieved an accuracy of 0.85, with a precision of 0.70, recall of 0.75, F1-score of 0.72, and an area under the ROC curve (AUC) of 0.69. These findings suggest that Hier-ViT can effectively classify dysphagia severity in LMI patients using early MRI, offering a potential tool for early risk stratification. While the model shows a high accuracy, the modest AUC suggests that further refinement and multi-modal integration are necessary to improve its discriminative power in imbalanced clinical datasets.