CHEST MR IMAGE SEGMENTATION USING A LEVEL SET METHOD BASED ON DISCRIMINANT CRITERIA(International Workshop on Advanced Image Technology 2007)

概要

論文の詳細を見る
Segmentation is a important problem for medical imaging analysis and visualization. Level set method has been widely applied to medical image segmentation. However, speed function of the level set method has a problem that the boundary with a weak edge cannot be properly determined. Moreover, it is necessary to extract contour of lungs and contour of blood vessel separately. We apply two ideas to segment a Chest MR image. One is use of discriminant criteria. Another is use of two kinds of curves which adapt to contour of lung and contour of blood vessels. By using these two ideas, two kinds of curves are controlled according to a different speed function automatically. We apply our method to extract air region in lung. Experimental results show that the proposed method extracts contour of lung and blood vessels effectively, in the case of the initial curve enclosing blood vessels.
社団法人電子情報通信学会の論文
2007-01-03