Paper: | FR-PM-PS1.4 |
Session: | Atlas and Model Based Image Segmentation |
Time: | Friday, April 7, 13:30 - 14:50 |
Presentation: |
Poster
|
Title: |
Sketch Initialized Snakes for Rapid, Accurate, and Repeatable Interactive Medical Image Segmentation |
Authors: |
Tim McInerney; Ryerson University | | |
| M. Reza Akhavan Sharif; Ryerson University | | |
Abstract: |
We combine a pen and pressure-sensitive tablet input device, and a sketch-based user initialization process, with a general subdivision-curve Snake to create an intuitive, fast, accurate, interactive model-based segmentation method. Using the pen input device, the Snake is quickly and precisely initialized with a series of sketch lines such that it is extremely close to the position and shape of the target object boundary, making the Snake’s task much simpler and hence more likely to succeed in noisy images with minimal user editing. The user may also use pen pressure levels to impart knowledge of object edge strength to the model, and the low degree-of-freedom subdivision curve Snake provides powerful control and editing capabilities. We apply our Snake to the segmentation of several 2D medical images. |