Towards complete 3D segmentation and visualization of diseased lung lobes
Date
2011
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The ultimate solution for effective diagnosis and treatment of lung diseases is the research and development of a computer-based system, which automatically converts a stack of two-dimensional (2D) computed tomographic images of lungs into a three-dimensional (3D) visualization. Current segmentation systems are unsuitable for clinical use due to one or more of the following constraints: (1) lack of automation, (2) inability to deal with multiple lung diseases and (3) incompleteness of the segmentation in the 3D space. The objective of this thesis is to research and develop a prototype for 3D segmentation and visualization of diseased lung lobes that simultaneously addresses the above three constraints. The prototype employs novel methods of texture analysis, fissure region analysis and surface modeling based on ridge regression to segment diseased lung lobes. Combined with a custom developed application programming interface in the 3D visualization software Amira, the prototype automatically converts a stack of 2D CT images into a 3D visualization of lungs with a few clicks. Tested on CT image stacks from 24 patients with a variety of lung diseases (emphysema, small cell lung cancer, large cell lung cancer, and bronchiectasis), the system produced root-mean squared errors of 2.21 ± 1.21 mm, 2.51 ± 1.36 mm, and 2.38 ± 1.27 mm in segmenting the left oblique, right oblique and right horizontal fissures, respectively ( compared to manual segmentation of the same fissures). These results indicate the feasibility of developing a clinical system for 3D segmentation and visualization of diseased lung lobes.
Description
Bibliography: p. 118-127.
A few pages are in colour.
A few pages are in colour.
Keywords
Citation
Wei, Q. (2011). Towards complete 3D segmentation and visualization of diseased lung lobes (Doctoral thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/3952