Intuitive and Smart Editing of Three-Dimensional Geometric Heart Valve Apparatus Models from Cardiac CT Data
Lades F, Wels M, Steidl S, Sühling M (2015)
Publication Language: English
Publication Type: Conference contribution
Publication year: 2015
Pages Range: 29-36
Conference Proceedings Title: The 2nd Interactive Medical Image Computing Workshop
Event location: Munich, Germany
URI: http://www5.informatik.uni-erlangen.de/Forschung/Publikationen/2015/Lades15-IAS.pdf
Abstract
In cases of aortic valve diseases, diagnosis and subsequent interventional planning are highly facilitated by deriving exact three-dimensional geometric models of a patient's aortic valve apparatus from Cardiac Computed Tomography Angiography data. Fully-automatic approaches to do so however lack in absolute reliability and manual editing of the initially detected geometric model is often required. We therefore present an interactive editing method for this scenario - in particular for editing the aortic root model - based on the As-Rigid-As-Possible (ARAP) surface modeling paradigm, which allows efficient, robust, intuitive and physically plausible deformations of three-dimensional geometric models. The user constrains a model's surface by setting and moving handles - the so-called constraints - while the remaining, and only the remaining, free part of the surface is deformed automatically in a way such that the global shape is preserved in real-time. We extended the classical ARAP approach for our scenario by an energy smoothness regularization to overcome non-smooth artifacts at constrained positions. We furthermore incorporated the use of image evidence-based cues in the interactive work ow such that handles can be made "snap" into visible 3-D surface indicators. We evaluated our method in a user study regarding intuitiveness, achievable accuracy, inter-user variability, required time and robustness. The participants started with an initial average mesh-to-mesh surface error of 1.65 mm and achieved after 50 mouse clicks on average and less than 3.5 minutes an average error of 0.48 mm with respect to an expert-de ned ground truth. The inter-user variability was 0.43 mm.
Authors with CRIS profile
Involved external institutions
How to cite
APA:
Lades, F., Wels, M., Steidl, S., & Sühling, M. (2015). Intuitive and Smart Editing of Three-Dimensional Geometric Heart Valve Apparatus Models from Cardiac CT Data. In The 2nd Interactive Medical Image Computing Workshop (pp. 29-36). Munich, Germany, DE.
MLA:
Lades, Félix, et al. "Intuitive and Smart Editing of Three-Dimensional Geometric Heart Valve Apparatus Models from Cardiac CT Data." Proceedings of the MICCAI IMIC 2015, Munich, Germany 2015. 29-36.
BibTeX: Download