Related Experiment Video
Updated: May 26, 2026
Three-Dimensional Reconstruction for the Whole Lung with Early Multiple Pulmonary Nodules
Published on: October 13, 2023
Development of a model to predict high-risk histopathological features in early T-stage lung adenocarcinoma using
Suzune Tsukamoto1,2, Sho Koyasu1, Masatsugu Hamaji3,4
1Department of Diagnostic Imaging and Nuclear Medicine, Graduate School of Medicine, Kyoto University, Kyoto, Japan.
Background:
There has been growing interest in the limited resection of early-stage lung cancer. Imaging-based preoperative risk stratification is expected to improve outcomes, but has not yet been well established. This study aimed to develop an imaging-based model for predicting high-risk histopathological features in early T-stage lung adenocarcinoma to identify low-risk patients.
Methods:
We retrospectively enrolled patients with cTis or cT1 N0M0 resected adenocarcinomas who underwent thin-slice computed tomography (CT) and 18F-fluorodeoxyglucose positron emission tomography/CT (FDG-PET/CT) at two institutions (Kyoto University Hospital as the training set and Hyogo Prefectural Amagasaki General Medical Center as the external test set). The thin-slice CT parameters analyzed included volume-based consolidation-to-tumor ratio (vCTR), diameter-based CTR (dCTR), tumor volume (TV), and doubling time (DT); while the FDG-PET/CT parameters comprised maximum standardized-uptake value (SUVmax), SUVmean, metabolic tumor volume (MTV), and total lesion glycolysis (TLG). Histopathological risk-positive lesions were defined as positive for pleural, vessel, or lymphatic invasion; spread through air space (STAS); and International Association for the Study of Lung Cancer (IASLC) grade 3. Logistic regression was used to develop predictive models, which were then tested on the external test set.
Results:
We analyzed 190 lesions in 174 patients for the training set, and 34 lesions in 28 patients for the external test set. The parameters vCTR and SUVmax showed the highest area under the receiver operating characteristics curve (AUC) values for risk-positive lesions (vCTR >58%: 0.83; SUVmax >2.0: 0.87) as independent predictors. The AUCs of our model (predictors: vCTR, SUVmax) were 0.87 and 0.85 in the training and external test sets, respectively.
Conclusions:
We developed and externally validated an imaging-based model for predicting high-risk histopathological features in early T-stage lung adenocarcinoma. This model may prove valuable for identifying suitable candidate patients for limited surgery.
