Title :
Snake-based approach for segmenting pedicles in radiographs and its application in three-dimensional vertebrae reconstruction
Author :
Zhang, Junhua ; Shi, Xinling ; Wang, Yuanyuan ; Lv, Liang ; Wu, Jun
Author_Institution :
Dept. of Electron. Eng., Yunnan Univ., Kunming, China
Abstract :
A gradient vector flow (GVF) snake based method was proposed for pedicle segmentation in vertebral radiographs. Since pedicles were oval-shaped, the elliptical shape prior was used to constrain the evolution of the GVF snake. From segmented pedicles, some landmarks were automatically identified for 3D stereoradiographic reconstruction of vertebrae to reduce the observer variability. Ten radiographs including 260 pedicles were used to evaluate the segmentations. Results demonstrated that the distance between contours manually delineated by the user and those segmented by the proposed algorithm was far less than the distance resulted from the traditional GVF snake. The 3D reconstruction variance was reduced by using the landmarks obtained from the segmented pedicles. These results indicated that utilizing the elliptical shape prior improved the GVF snake for segmenting pedicles in radiographs, and the proposed method might be a useful preprocessing tool for 3D stereoradiographic reconstruction.
Keywords :
computer graphics; image reconstruction; image segmentation; radiography; 3D stereoradiographic reconstruction; elliptical shape; gradient vector flow; pedicle segmention; snake based method; snake-based approach; three-dimensional vertebrae reconstruction; vertebral radiographs; Computational modeling; Force; Image reconstruction; Image segmentation; Radiography; Shape; Three dimensional displays; Pedicle segmentation; elliptical shape prior; gradient vector flow (GVF) snake; radiograph; three-dimensional (3D) reconstruction;
Conference_Titel :
Image Processing (ICIP), 2010 17th IEEE International Conference on
Conference_Location :
Hong Kong
Print_ISBN :
978-1-4244-7992-4
Electronic_ISBN :
1522-4880
DOI :
10.1109/ICIP.2010.5652596