Identification of the optimal threshold for predicting the infiltration degree of T1-stage lung adenocarcinoma using solid component volume and three-dimensional consolidation-to-tumor ratio in threshold segmentation.

Abstract

Background: Predicting the invasiveness of pulmonary nodules when early-stage lung cancer is suspected is a clinical challenge. This study aimed to determine the optimal computed tomography (CT) threshold values for predicting the invasiveness of T1-stage lung adenocarcinoma. This was achieved using the solid component volume and three-dimensional consolidation-to-tumor ratio (3D CTR) via threshold segmentation.

Methods: A retrospective study was conducted, involving 1,056 patients with 1,179 pulmonary nodules verified by postoperative pathology. These cases were sourced from two different centers. The patients were divided into two groups: the pre-invasive group, comprising atypical adenomatous hyperplasia (AAH) and adenocarcinoma in situ (AIS), and the invasive group, including minimally invasive adenocarcinoma (MIA) and invasive adenocarcinoma (IAC). Seven different CT threshold settings [-550, -450, -350, -250, -150, -50, 0 Hounsfield unit (HU)] were used, and the solid component volume was calculated; 3D CTR was determined using the threshold segmentation method and the differences between the two groups were analyzed. We plotted the receiver operating characteristic (ROC) curves to evaluate the effectiveness of predicting the invasiveness of T1-stage lung adenocarcinoma. Based on the analysis of the ROC curves, the optimal threshold was determined, and the corresponding optimal cut-off value was calculated.

Results: The optimal predictive efficacy for evaluating the invasiveness of stage T1 lung adenocarcinoma was achieved with a -350 HU CT threshold. The predictive performance for the invasiveness of T1-stage lung adenocarcinoma was optimal. The area under the ROC curve (AUC) with its 95% confidence interval (CI) for the solid component volume was 0.855 (0.834-0.876), and for the 3D CTR, it was 0.823 (0.799-0.847). The optimal cutoff point for the solid component volume was 45.5 mm³, and 10.85% for 3D CTR.

Conclusions: Regardless of the CT threshold setting, the solid component volume and 3D CTR calculated based on the threshold segmentation method were demonstrated to be stable predictive factors that significantly contributed to the assessment of the invasiveness of T1-stage lung adenocarcinoma. The optimal predictive performance was achieved when the CT threshold was set to -350 HU. A solid component volume exceeding 45.5 mm³ or a 3D CTR greater than 10.85% indicated a higher likelihood of MIA or IAC.