Fast and Interactive 3D-Segmentation of Medical Volume Data

The segmentation of tomographic image data is an important prerequisite for a meaningful vi-sualization in medicine. To circumvent diicul-ties related to automatic approaches, we suggest interactive segmentation which includes the knowledge of a user more eeciently. Using intelligent scissors, as presented in 1] for the 2D case, we suggest 3D lters for the calculation of the cost matrix, an automatic procedure which propagates contours to adjacent slices and three communicating 2D displays for convenient delin-eation of contours in volume data. Additionally, we provide volume growing based on a statistical approach, as presented in 2, 3], allowing to select coherent subvolumes interactively. For the immediate evaluation we introduce additional 3D displays for the visualization of polygonal and vol-umetric representations of the segmentation results .