获取途径

作者

Jia Guo, Weikai Sun, Tongyu Wang, Jianguo Miao, Pei Nie, Yong Huang, Wenjian Xu

作者单位

Department of Radiology, The Affiliated Hospital of Qingdao University, Qingdao, China (J.G., T.W., P.N., W.X.). ² Department of Radiology, Qilu Hospital of Shandong University, Jinan, Shandong, China (W.S.). ³ Department of Radiology, The Affiliated Hospital of Qingdao University, Qingdao, China (J.G., T.W., P.N., W.X.). ⁴ School of Engineering Medicine, Beihang University, Beijing, China (J.M.). ⁵ Department of Radiology, Shandong Cancer Hospital and Institute,Shandong First Medical University and Shandong Academy of Medical Sciences, China (J.G.,Y.H.). ⁶ Department of Radiology, The Affiliated Hospital of Qingdao University, Qingdao, China (J.G., T.W., P.N., W.X.). Electronic address: wjxu2021@qdu.edu.cn. Affiliations ¹ Department of Radiology, Shandong Cancer Hospital and Institute,Shandong First Medical University and Shandong Academy of Medical Sciences, China (J.G.,Y.H.)

刊名

Academic radiology

年份

2026

ISSN

1878-4046

关键词

Bone metastasis Deep learning Lung adenocarcinoma Prognostic models.

摘要

Rationale and objectives: Preoperatively identifying patients at high risk of bone metastasis remains challenging in resectable lung adenocarcinoma , limiting early risk-adapted surveillance. We aimed to develop and validate an interpretable CT-based deep learning model-DL bone metastasis-free survival prediction signatures -to predict BMFS and provide time-dependent BM risk probabilities across follow-up. Materials and methods: In this retrospective multicohort study, 1042 patients with preo...更多

Rationale and objectives: Preoperatively identifying patients at high risk of bone metastasis remains challenging in resectable lung adenocarcinoma , limiting early risk-adapted surveillance. We aimed to develop and validate an interpretable CT-based deep learning model-DL bone metastasis-free survival prediction signatures -to predict BMFS and provide time-dependent BM risk probabilities across follow-up. Materials and methods: In this retrospective multicohort study, 1042 patients with preoperative computed tomography were included . DBPs were evaluated using the area under the receiver operating characteristic curve and Brier score. Interpretability was assessed by associating CT-derived DL features with histopathologic risk factors through unsupervised clustering and gradient boosting with SHapley Additive exPlanations . Results: DBPs predicted BMFS with AUCs of 0.822 and a Brier score of 0.08 in internal validation, and an AUC of 0.800 and a Brier score of 0.10 in external validation. Unsupervised clusters derived from CT-based DL features were significantly associated with histopathologic risk factors , except epidermal growth factor receptor mutation status . DL features also showed moderate discrimination for vascular invasion and visceral pleural invasion , with AUCs of 0.738 and 0.708 , respectively, supporting the biological interpretability of the imaging signature. Conclusion: An interpretable CT-based DL model enables preoperative BMFS prediction in resectable LUAD, and supports interpretability by relating imaging-derived representations to established histopathologic aggressiveness factors.收起

文献类型

期刊

浏览量

3