Related Experiment Video
Updated: Jul 6, 2026
Author Spotlight: Advancing 3D Modeling for Enhanced Diagnosis and Treatment of Pulmonary Nodules in Early-Stage Lung Cancer
Published on: October 13, 2023
Prediction of pulmonary complications post-lobectomy or -sub-lobectomy in lung cancer using artificial
Wanyi Zheng1, Borong Tang2, Yunjing Xue1
1Department of Radiology, Fujian Medical University Union Hospital, Fuzhou, China.
Background:
Several studies have demonstrated that artificial intelligence-based computed tomography (AI-CT) lung volume index can predict complications after lobectomy. However, previous studies have focused more on AI-CT results at the whole lung, left and right lung, and lobar levels. This study explored the value of the predicted postoperative pulmonary function values derived from preoperative chest AI-CT in predicting pulmonary complications (PCs) after lobectomy or sub-lobectomy in lung cancer.
Methods:
A total of 79 patients with lung cancer who underwent thoracoscopic lobectomy or sub-lobectomy were enrolled. Chest computed tomography (CT) and pulmonary function tests (PFTs) were performed one week before surgery. The patients were divided into PC+ (n=16) and PC- (n=63) groups according to whether PC occurred within 30 days after the operation. The total functional lung volume (TFLV) and resected functional lung volume (RFLV) were calculated using the AI-CT method. The predicted postoperative pulmonary function indexes, including total lung capacity (TLC), forced expiratory volume in the first second (FEV1), forced vital capacity (FVC), and diffusion capacity for carbon monoxide (DLCO), were calculated using the AI-CT and anatomical segmentation (AS) methods. PC-related factors were identified by univariate analysis to construct clinical, PFT, AS, AI-CT, clinical-PFT, clinical-AS, and clinical-AI-CT models. Sub-group analysis was performed based on the surgery procedure.
Results:
Age, number of removed lung segments, and RFLV/TFLV in the PC+ group were higher than those in the PC- group (P<0.05); FEV1, FVCAI-CT, FEV1AI-CT, FVCAS, and FEV1AS were lower in the PC+ group than those in PC- group (P <0.05). Lobectomy led to more complications than sub-lobectomy [univariate analysis odds ratio (OR) (95% confidence interval, 95% CI): 0.259 (0.082-0.815), P<0.05]. FEV1AI-CT [OR: 0.077, P=0.026; relative risk (RR): 0.261, P<0.001] was an independent predictor. Comparison of the areas under the curve: clinical-AI-CT (0.823) > AI-CT (0.807) > clinical-PFT (0.806) > clinical-AS (0.801) > AS (0.767) > clinical (0.757) > PFT (0.756) model. In sub-group analysis, only FEV1AI-CT showed statistical differences (P<0.05) between PC+ and PC- in both lobectomy and sub-lobectomy subgroups.
Conclusions:
FEV1AI-CT is a potential independent predictor of PC after lung lobectomy or sub-lobectomy in lung cancer.
