Interobserver Variability in Manual Versus Semi-Automatic CT Assessments of Small Lung Nodule Diameter and Volume

Area of Science:

• Radiology
• Medical Imaging Analysis
• Thoracic Imaging

Background:

• Assessing small lung nodules on computed tomography (CT) scans is crucial for early cancer detection.
• Interobserver variability in nodule measurements can impact diagnostic accuracy and patient management.
• Current manual measurement techniques may be subject to significant observer bias.

Purpose of the Study:

• To compare the interobserver variability of semi-automatic diameter and volumetric measurements against manual diameter measurements for small lung nodules.
• To evaluate the effectiveness of semi-automatic techniques in reducing measurement discrepancies.
• To inform best practices in lung nodule characterization using CT imaging.

Main Methods:

• Retrospective analysis of CT thorax examinations with noncalcified solid nodules (3-10 mm).
• Three experienced radiologists performed manual diameter, semi-automatic diameter, and volumetric measurements.
• Intermeasurement agreement assessed using Spearman's correlation and Bland-Altman plots.

Main Results:

• Semi-automatic diameter and volume measurements demonstrated significantly stronger agreement between reviewers (avg. r = 0.97 and 0.99) than manual measurements (avg. r = 0.91).
• Semi-automatic diameter measurements outperformed manual methods in all comparisons.
• Volumetric measurements showed superior agreement in two out of three comparisons.

Conclusions:

• Semi-automatic measurement techniques substantially decrease interobserver variability for small solid lung nodules.
• Both semi-automatic diameter and volumetric assessments offer improvements over traditional manual methods.
• Standardized training may further reduce observer variability in semi-automatic nodule analysis.