Conference proceeding
3D segmentation of non-isolated pulmonary nodules in high resolution CT images
Proceedings of SPIE, Vol.5747(1), pp.1438-1445
Medical Imaging 2005: Image Processing
04/29/2005
DOI: 10.1117/12.594938
Abstract
The purpose of this study is to develop a computer-aided diagnosis (CAD) system to segment small size non-isolated pulmonary nodules in high resolution helical CT scans. A new automated method of segmenting juxtapleural nodules was proposed, in which a quadric surface fitting procedure was used to create a boundary between a juxtapleural nodule and its neighboring pleural surface. Experiments on some real CT nodule data showed that this method was able to yield results that reflect the local shape of the pleural surface. Additionally, a scheme based on parametrically deformable geometric model was developed to deal with the problem of segmenting nodules attached to vessels. A vessel segment connected to a nodule was modeled using superquadrics with parametric deformations. The boundary between a vascularized nodule and the attached vessels can be recovered by finding the deformed superquadrics which approximates the vessels. Gradient descent scheme was utilized to optimize the parameters of the superquadrics. Simple experiments on synthetic data showed this scheme is promising.
Details
- Title: Subtitle
- 3D segmentation of non-isolated pulmonary nodules in high resolution CT images
- Creators
- Xiangwei Zhang - University of IowaGeoffrey McLennan - University of IowaEric A Hoffman - University of IowaMilan Sonka - University of Iowa
- Resource Type
- Conference proceeding
- Publication Details
- Proceedings of SPIE, Vol.5747(1), pp.1438-1445
- Conference
- Medical Imaging 2005: Image Processing
- DOI
- 10.1117/12.594938
- ISSN
- 0277-786X
- Language
- English
- Date published
- 04/29/2005
- Academic Unit
- Roy J. Carver Department of Biomedical Engineering; Radiology; Electrical and Computer Engineering; Radiation Oncology; Fraternal Order of Eagles Diabetes Research Center; Injury Prevention Research Center; Internal Medicine; Ophthalmology and Visual Sciences
- Record Identifier
- 9984186592802771
Metrics
14 Record Views