Book chapter
Segmentation of Lungs with Interstitial Lung Disease in CT Scans: A TV-L1 Based Texture Analysis Approach
Advances in Visual Computing, pp.511-520
Lecture Notes in Computer Science, Springer International Publishing
2014
DOI: 10.1007/978-3-319-14249-4_48
Abstract
Lung segmentation methods are important for automated lung image analysis tasks such as quantification of lung diseases. In this paper, we describe a method for segmentation of lungs with interstitial lung disease (ILD). In thoracic CT scans, such lungs are characterized by the presence of texture patterns like honeycombing, which makes lung segmentation difficult. We employ a 3D total variation L1 (TV-L1) based texture analysis approach to extract these patterns and attenuate the density of the corresponding voxels in the CT scan. The modified CT scan is then utilized as input to an existing 3D robust active shape model based lung segmentation method. The proposed method was evaluated on 77 CT scans of lungs with and without ILD. On cases with ILD, our method obtained an average volumetric overlap of 0.95±0.02, which was statistically significantly better than two other approaches. The TV-L1 texture analysis utilizes GPUs, making our method fast.
Details
- Title: Subtitle
- Segmentation of Lungs with Interstitial Lung Disease in CT Scans: A TV-L1 Based Texture Analysis Approach
- Creators
- Gurman Gill - The Iowa Institute for Biomedical Imaging, The University of Iowa, USAReinhard R Beichel - The Iowa Institute for Biomedical Imaging, The University of Iowa, USA
- Resource Type
- Book chapter
- Publication Details
- Advances in Visual Computing, pp.511-520
- Publisher
- Springer International Publishing; Cham
- Series
- Lecture Notes in Computer Science
- DOI
- 10.1007/978-3-319-14249-4_48
- eISSN
- 1611-3349
- ISSN
- 0302-9743
- Language
- English
- Date published
- 2014
- Academic Unit
- Electrical and Computer Engineering
- Record Identifier
- 9984083210702771
Metrics
13 Record Views