Detection of convergent folds using curvilinear objects in double contrast X-ray image of stomach

摘要

This paper proposes a new method to detect convergent folds by extracting of continuous curves and tangential directions. Folds are the line convex regions that have various contrasts and complex line-structures. Then we use the line-convergence vector field model and line-object tracking method in an object space for understanding these structures. This model and method enable easy and stabile understanding of each line structure including tangential direction. Tangential directions and end points of curves can indicate convergent folds for us. The effectiveness of the method is shown by comparing the experimental results with the sketch by a radiologist.