Exploring and verifying key thin-section computed tomography features for accurately differentiating granulomas and peripheral lung cancers.

Background: Granulomas were frequently misdiagnosed as peripheral lung cancers (PLCs) due to their similarities in imaging findings. This study aimed to establish a classification system based on thin-section computed tomography (TSCT) features for distinguishing granulomas from PLCs.

Methods: From January 2012 to November 2023, 561 granulomas and 561 size-matched PLCs manifested as solid nodules (SNs) were retrospectively enrolled. Their TSCT features were evaluated and compared. Based on single or a combination of multiple features, a classification system comprising multiple distinct types with different features was established for differentiation.

Results: Lesions were classified into eight types. Types I (nodules with satellite lesions) [28.3% 2.1%; positive predictive value (PPV), 0.93], II (nodules with halo sign and/or calcification) (16.4% 3.0%; PPV, 0.84), III (nodules with a special shape) (7.7% 0.5%; PPV, 0.93), and IV (nodules with ill-defined boundary) (8.6% 4.3%; PPV, 0.67) were more common in granulomas than in PLCs (each P<0.05); among other well-defined nodules, type V (non-lobulated nodule with smooth margin) (16.0% 4.6%; PPV, 0.78) was common in granulomas (P<0.001), while types VI (non-lobulated nodule with coarse margin) (7.8% 14.1%; PPV, 0.64), VII (lobulated nodule with vacuole sign) (1.2% 13.5%; PPV, 0.92), and VIII (lobulated nodule without vacuole sign) (14.8% 57.8%; PPV, 0.80) were more frequent in PLCs (each P≤0.001). Receiver operating characteristic (ROC) curve analysis showed that types I, II, III, V, VII, and VIII were more effective in differentiation (each P<0.05).

Conclusions: In differentiating pulmonary SNs, those with TSCT features manifested as type I, II, III, or V had a higher possibility of granulomas.