Registration of CT and MR Abdominal Images
| Autor: | Min-Shen Liu, 劉明賢 |
|---|---|
| Rok vydání: | 1996 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 84 Image registration is the process of determining a point-to- point correspondence between two images of the same object. This is a crucial first step for fusing two image data sets to obtain an integrated image display. Feature base methods are often used in image registration, but they have trouble in registering abdominal images, because it is difficult to extract geometrical invariance features from abdominal images, since these images are too complex and usually distorted. Registration using correlation technique needn't image features, so we tried to use correlation technique to register abdominal images. For using correlation tech- niques, we transform CT and MR images to classified images, which are two similar images. Since calculating correlation is very time-consuming, we use a more specific method to match CT and MR images first. We get scale and rough translation between those two images by calculating body area and central point from extracted body contour. Then, K- mean clustering method was used to classify tis- sues in CT image. And fuzzy inference method was used to classify tissues in MR image. The relation table con- taining the relations between classes in the two classi- fied images was established for counting correlation. Then we count the correlation between two classified images to find which translation and rotation degree derives the highest correlation. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|