This paper presents a new deformable model for object detection, based on charged particle dynamics and geometric contour propagation, applied to the segmentation of the left ventricle of the heart in SPECT images. The model detects object boundaries with a charged active contour that propagates under the influence of Lorentz forces in an image-based electrostatic field. We find positions of vector field divergence for automatic initialisation of the contour. Experimental results on a dataset of 160 images are evaluated against hand-labelled groundtruth data.
|Translated title of the contribution
|A Charged Contour Model for Cardiac SPECT Segmentation
|Title of host publication
|Published - Jul 2006