Related Experiment Video
Updated: Mar 22, 2026
05:56
Author Spotlight: Enhancing Diagnostic Strategies and Biomarker Development for Comprehensive Lung Function Analysis
Published on: August 9, 2024
2.8K
Inter-observer Variability Analysis of Automatic Lung Delineation in Normal and Disease Patients
Luca Saba1, Joel C M Than2, Norliza M Noor3
1Azienda Ospedaliero Universitaria (A.O.U.) di Cagliari - Polo di Monserrato, Università di Cagliari, s.s. 554 Monserrato, Cagliari, 09045, Italy.
Journal of Medical Systems
|April 27, 2016
Summary
This study analyzed observer variability in automated lung delineation using High Resolution Computed Tomography (HRCT) images. Results show minimal differences between observers, validating the automated system
Area of Science:
- Medical Imaging Analysis
- Radiology
- Biomedical Engineering
Background:
- Automated medical systems require human interaction, leading to data variability.
- Regulatory approval necessitates analysis of inter-observer and intra-observer variability.
- Lung delineation accuracy is crucial for automated medical system validation.
Purpose of the Study:
- To assess inter-observer variability in an automated lung delineation system.
- To validate the accuracy of automated lung delineation against manual tracings.
- To analyze the impact of observer experience on delineation accuracy.
Main Methods:
- Three observers manually delineated lung borders on High Resolution Computed Tomography (HRCT) images using ImgTracer™ software.
- Data analyzed using D'Agostino-Pearson test, ANOVA, Dice Similarity Coefficient (DSC), Jaccard Index (JI), and Hausdorff Distance (HD).
- Statistical comparisons included regression plots, Bland-Altman plots, T-tests, Mann-Whitney, and Chi-Squared tests.
Main Results:
- All observers' tracings showed normal distribution (P > 0.05).
- ANOVA tests revealed high P-values (>0.81) for both lungs, indicating minimal variability.
- Observer Deterioration Factor (ODF) showed less than 10% difference for the less experienced observer.
Conclusions:
- The automated lung delineation system demonstrates acceptable accuracy and minimal inter-observer variability.
- Observer experience has a limited impact on the system's performance.
- The system successfully identified diseased lungs as smaller than normal lungs.
