Abstract

Segmentation of the left ventricle(LV) from CT imagery is frequently required in evaluating the ventricle volume, mass, and ejection fraction. A key step in the segmentation is to globally locate the LV in a CT image. In this work, we propose an automatic method for localizing the LV from CT images by formulating the problem as surface segmentation on the heart surface. In this formulation, global geometric features are naturally integrated into the segmentation framework. The robustness of this method has been demonstrated by experimental results from 34 CT images.

© 2013 Optical Society of America