Clinical data of patients who underwent pulmonary nodule resection in our Hospital between January 2016 and April 2024 were retrospectively analyzed. Inclusion criteria were: (1) CT imaging showing a GGN with a maximum diameter of ≤ 3 cm; (2) surgical pulmonary nodule resection with postoperative pathology analysis; (3) CT plain scan and contrast enhanced scan before surgery. Exclusion criteria were 1) pulmonary GGNs > 3cm and 2) a different pathological diagnosis from adenocarcinoma.

The qualitative parameters analyzed included:

1. Nodule type (pure GGN or mixed GGN)
2. Morphology (rounded/round-like or irregular)
3. Type of feeding vessel (branches of the pulmonary arteries or veins)
4. Presence of the bronchus sign
5. Presence of a pleural tag
6. Presence of bubble-like lucency

In addition, five quantitative parameters were assessed on CT imaging:

1. Nodule size: Defined as the largest diameter measured among the axial, sagittal, and coronal planes.
2. Maximum CT attenuation: Measured in Hounsfield Units (HU) for the ground-glass component of nodules and the solid component of part-solid nodules, in both non-contrast and arterial-phase scans.
3. Lung parenchyma CT attenuation: Measured in HU within a normal lung region located 10 mm from the nodule’s edge, excluding blood vessels and bronchi.
4. Difference CT attenuation: Defined as the difference between the maximum CT attenuation and the lung parenchyma CT attenuation.
5. Relative attenuation: Calculated as the ratio of lung parenchyma CT attenuation to maximum CT attenuation.

Difference CT attenuation and relative attenuation were calculated in accordance with the study by L. Qi et al. [4]. Measurements were performed on thin-section CT images in both plain and contrast-enhanced scans, using a 20 mm² region of interest (ROI) while carefully avoiding blood vessels and bronchial structures.

To evaluate the predictive value of these parameters, GGNs were classified into two groups based on pathological diagnosis: the Non-IAC Group (including AIS, AAH and MIA) and the IAC Group. ROC curve analysis was performed for each quantitative parameter. Qualitative and quantitative data were compared across different pathological subtypes. Additionally, binary logistic regression was applied to identify independent predictors of GGN invasiveness.