Related Experiment Video
Updated: May 2, 2026
10:26
Author Spotlight: A 3D Digital Model for the Diagnosis and Treatment of Pulmonary Nodules
Published on: May 19, 2023
1.9K
Application of computer-aided detection for NCCN-based follow-up recommendation in subsolid nodules: Effect on
Wu Quanyang1, Zhou Lina1, Huang Yao1
1Department of Diagnostic Radiology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China.
Cancer Medicine
|February 13, 2024
Summary
Computer-aided detection (CAD) significantly improves agreement on managing subsolid nodules among radiologists. This tool reduces discrepancies and enhances lung cancer detection sensitivity, aiding clinical decisions.
Area of Science:
- Radiology
- Medical Imaging
- Oncology
Background:
- Observer variability impacts nodule assessment reliability in clinical practice.
- Computer-aided detection (CAD) aims to reduce this variability for improved diagnostic accuracy.
Purpose of the Study:
- To evaluate the impact of CAD on inter-observer agreement for subsolid nodule follow-up management.
- To assess CAD's effect on reducing management discrepancies and improving detection sensitivity.
Main Methods:
- Five observers independently assessed 60 subsolid nodule cases using low-dose CT scans.
- Follow-up management strategies were assigned per NCCN guidelines, with and without CAD assistance.
- Inter-observer agreement was measured using Cohen's kappa and Fleiss kappa statistics.
Main Results:
- Manual assessment showed moderate agreement (Fleiss kappa = 0.437).
- CAD use significantly improved agreement to a substantial level (Fleiss kappa = 0.623).
- CAD reduced major discrepancies from 27.5% to 15.8% and substantial discrepancies from 4.8% to 1.5% (p < 0.01).
- CAD increased sensitivity for detecting part-solid lung nodules (82.6% vs. 92.2%, p < 0.05).
Conclusions:
- CAD substantially enhances inter-observer agreement in subsolid nodule management.
- CAD application reduces significant management discrepancies and improves lung cancer detection sensitivity.
