Related Experiment Video
Updated: Jul 2, 2026
Lung CT Segmentation to Identify Consolidations and Ground Glass Areas for Quantitative Assesment of SARS-CoV Pneumonia
Published on: December 19, 2020
Development of an automatic classification system for differentiation of obstructive lung disease using HRCT
Namkug Kim1, Joon Beom Seo, Youngjoo Lee
1Department of Radiology and Research Institute of Radiology, University of Ulsan College of Medicine, Asan Medical Center, 388-1, Pungnap2-dong, Songpa-gu, Seoul, 138-736, Republic of Korea.
Adding shape features to texture analysis significantly improves the detection of obstructive lung diseases using high-resolution computerized tomography (HRCT) scans. This enhancement aids in classifying conditions like emphysema and bronchiolitis obliterans more accurately.
Area of Science:
- Medical Imaging
- Radiology
- Computer-Aided Diagnosis
Background:
- Obstructive lung diseases present diagnostic challenges.
- High-resolution computerized tomography (HRCT) is crucial for lung imaging.
- Accurate classification of lung diseases like emphysema and bronchiolitis obliterans is vital.
Purpose of the Study:
- To introduce novel shape features for HRCT image analysis.
- To optimize classification algorithms for improved differentiation of obstructive lung diseases.
- To evaluate the impact of shape features on classification performance.
Main Methods:
- Analysis of 265 HRCT images from 82 subjects.
- Inclusion of 11 shape features alongside 13 textural features.
- Implementation of Bayesian and Support Vector Machine (SVM) classifiers.
- Optimization of classifier parameters and region of interest (ROI) sizes (16x16, 32x32, 64x64 pixels).
- Utilized a five-fold cross-validation method.
Main Results:
- Incorporating shape features significantly improved overall sensitivity compared to texture features alone (up to 9.1% increase with SVM).
- Optimized SVM with selected features achieved high sensitivity (93.5 +/- 1.0%) at 64x64 pixel ROI size.
- Shape features demonstrated greater contribution to sensitivity in smaller ROIs.
Conclusions:
- Shape features enhance the classification performance of obstructive lung diseases on HRCT images.
- The combination of texture and shape features offers a more robust diagnostic approach.
- This method shows promise for automated and accurate lung disease diagnosis.
