Every year around 13 million people have a stroke, and approximately half of them suffer from swallowing difficulties. The prediction of the duration of poststroke dysphagia is an essential factor for doctors in treatment decisions. A stroke causing dysphagia will lead to dehydration and malnutrition. To avoid this, doctors must decide within 48 hours after the Stroke about enteral tube feeding. If dysphagia is likely to persist for more than 7 days, then nasogastric tube (NGT) is the recommended enteral tube feeding, whereas percutaneous endoscopic gastrostomy (PEG) placement if it is likely not to recover within 30 days.
There are various factors identified in the previous studies which prolong swallowing problems. The important factors include lesion location and volume, age, National Institutes of Health Stroke Scale (NIHSS) score, time from stroke onset, stroke severity on admission, initial impairment of oral intake, and initial risk aspiration.
One of the most important predictive variables for prolonging dysphagia is lesion location in the brain based on previous studies. In an earlier study, it was found that if you have a frontal operculum lesion, there is a higher chance of prolonging the swallowing problem. Determining the lesion location in the brain becomes difficult because of different imaging modalities, i.e., magnetic resonance imaging (MRI) and computed tomography (CT). Also, the lack of an experienced radiologist in remote areas or developing countries affects the quality of obtained results. In this research, we propose to use deep learning to develop a robust model to detect the lesion location with an accuracy equal to or more than an experienced radiologist. Additionally, we plan to explore other important factors that can be used in the deep learning-based model to predict the dysphagia recovery time.