Recognition of Barium Meal Track for Swallowing Imaging

Swallowing angiography video is a golden standard for evaluating the condition of patients with swallowing disorders. However, the current diagnosis of swallowing disorders depends on the experience of doctors and lacks quantitative analysis. In the current era of information technology, it is of great significance to use computers to process and quantitatively analyze swallowing radiography videos. Semantic segmentation is a commonly used method for image recognition and has been widely used in many fields. This article first produced a swallowing angiography video dataset with tag data, and then used various image processing methods to optimize image quality. Then, a semantic segmentation model is used to identify barium meal targets and perform discriminant tracking. Finally, the quantitative results were analyzed and a method for calculating the motion speed of barium meal was proposed, which can be used to assist doctors in diagnosis.