Classification of CT Pulmonary Opacities as Perifissural Nodules: Reader Variability

Area of Science:

• Radiology
• Pulmonary Medicine
• Oncology

Background:

• Pulmonary nodules are common incidental findings on CT scans.
• Differentiating benign perifissural nodules (PFNs) from malignant nodules is crucial for lung cancer screening.
• Interreader variability in nodule classification can impact patient management.

Purpose of the Study:

• To assess interreader variability in classifying pulmonary opacities as perifissural nodules (PFNs) on CT.
• To evaluate the reliability of radiologists in distinguishing PFNs from malignancies.

Main Methods:

• Retrospective analysis of CT scans from the National Lung Screening Trial.
• Six radiologists classified 359 nodules (5-10 mm, noncalcified, solid) as PFN, non-PFN, or not applicable.
• Statistical analysis of 316 nodules after excluding those classified as not applicable by multiple readers.

Main Results:

• The dataset included 22.2% cancers. Median PFN classification was 45.6%.
• Uniform classification of PFNs by all readers occurred in 17.7% of nodules (Fleiss κ = 0.50).
• PFNs were smaller, more often in lower lobes, and attached to fissures. 18.6% of cancers were misclassified as PFNs, with upper lobe location increasing misclassification risk.

Conclusions:

• Moderate interreader agreement exists for classifying perifissural nodules.
• Misclassified lung cancers as PFNs were rare (<2.5% of classifications).
• Omitting PFNs from follow-up, especially in upper lobes, could reduce unnecessary scans and misclassification rates.