Related Experiment Videos
Quantitative evaluation of a pulmonary contour segmentation algorithm in X-ray computed tomography images
Beatriz Sousa Santos1, Carlos Ferreira, José Silvestre Silva
1Serviço de Imagiologia, Hospitais da Universidade de Coimbra, Portugal.
Academic Radiology
|August 4, 2004
Summary
This study validates an automated pulmonary contour extraction method for thoracic CT scans. The method demonstrated accuracy comparable to radiologists, with lower variability than human observers, ensuring reliable pre-processing for image analysis.
Area of Science:
- Medical Imaging
- Radiology
- Image Processing
Background:
- Pulmonary contour extraction is crucial for automated thoracic CT analysis.
- Accurate segmentation is essential for reliable quantitative assessments.
Purpose of the Study:
- To quantitatively assess the performance of an automated pulmonary contour extraction method.
- To compare the method's accuracy and consistency against manual delineations by radiologists.
Main Methods:
- Statistical comparison of automated contours with manual contours from six radiologists on 30 thoracic CT images.
- Analysis of inter- and intra-observer variability using figures of merit.
Main Results:
- The automated method showed strong consistency across quality indexes.
- Inter-observer variability among radiologists was greater than the method's variability compared to individual radiologists.
- The method's performance as a pulmonary contour detector was similar to that of the radiologists.
Conclusions:
- The pulmonary contour extraction method's consistency and accuracy are adequate for most radiologist quantitative requirements.
- The developed evaluation methodology is robust and applicable to other segmentation scenarios.