講演抄録/キーワード |
講演名 |
2009-01-19 13:30
[ポスター講演]Automatic Prostate Segmentation using Shape and Gradient Information in MR Images Yu Jin Jang・○Helen Hong(Seoul Women's Univ.)・Hak Jong Lee・Sung Il Hwang(Seoul National Univ. Bundang Hospital) MI2008-96 |
抄録 |
(和) |
To detect the prostate cancer and assess the tumor stage, we present an automatic segmentation method that uses a prior knowledge of prostate shape to arrive at the exact prostate contour in MR images. First, statistical shape model of the prostate is constructed by using parametric mapping and PCA analysis. Then, the prostate contour is delineated by using Active Shape Model. To improve the segmentation of the apex in patients with prostatic hypertrophy, the two-dimensional correction method using gradient information is proposed. Our method has been applied to ten MR data sets. The result was evaluated using radiologist。ッs manual segmentation. The average distance difference error was 0.36mm and the overlapping volume ratio was 95%. |
(英) |
To detect the prostate cancer and assess the tumor stage, we present an automatic segmentation method that uses a prior knowledge of prostate shape to arrive at the exact prostate contour in MR images. First, statistical shape model of the prostate is constructed by using parametric mapping and PCA analysis. Then, the prostate contour is delineated by using Active Shape Model. To improve the segmentation of the apex in patients with prostatic hypertrophy, the two-dimensional correction method using gradient information is proposed. Our method has been applied to ten MR data sets. The result was evaluated using radiologist。ッs manual segmentation. The average distance difference error was 0.36mm and the overlapping volume ratio was 95%. |
キーワード |
(和) |
Prostate segmentation / Active Shape Model / MRI / / / / / |
(英) |
Prostate segmentation / Active Shape Model / MRI / / / / / |
文献情報 |
信学技報, vol. 108, no. 385, MI2008-96, pp. 173-174, 2009年1月. |
資料番号 |
MI2008-96 |
発行日 |
2009-01-12 (MI) |
ISSN |
Print edition: ISSN 0913-5685 Online edition: ISSN 2432-6380 |
著作権に ついて |
技術研究報告に掲載された論文の著作権は電子情報通信学会に帰属します.(許諾番号:10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
PDFダウンロード |
MI2008-96 |