Department of Radiology, Jiangmen Central Hospital, 23 Beijie Haibang Street, Jiangmen 529030, PR China.
School of Electronic Information and Automation, Guilin University of Aerospace Technology, 2 Jinji Road, Guilin 541000, PR China.
Eur J Radiol. 2023 Dec;169:111169. doi: 10.1016/j.ejrad.2023.111169. Epub 2023 Nov 7.
To develop and externally validate multiphase CT-based deep learning (DL) models for differentiating adrenal metastases from benign lesions.
This retrospective two-center study included 1146 adrenal lesions from 1059 patients who underwent multiphase CT scanning between January 2008 and March 2021. The study encompassed 564 surgically confirmed adenomas, along with 135 benign lesions and 447 metastases confirmed by observation. DL models based on multiphase CT images were developed, validated and tested. The diagnostic performance of the classification models, imaging phases and radiologists with or without DL were compared using accuracy (ACC) and receiver operating characteristic (ROC) curves. Integrated discrimination improvement (IDI) analysis and the DeLong test were used to compare the area under the curve (AUC) among models. Decision curve analysis (DCA) was used to assess the clinical usefulness of the predictive models.
The DL signature based on LASSO (DLSL) had a higher AUC than that of the other classification models (IDI > 0, P < 0.05). Furthermore, the precontrast phase (PCP)-based DLSL performed best in the independent external validation (AUC = 0.881, ACC = 82.9 %) and clinical test cohorts (AUC = 0.790, ACC = 70.4 %), outperforming DLSL based on the other single-phase or three-phase images (IDI > 0, P < 0.05). DCA demonstrated that PCP-based DLSL provided a higher net benefit (0.01-0.95). The diagnostic performance led to statistically significant improvements when radiologists incorporated the DL model, with the AUC improving by 0.056-0.159 and the ACC improving by 0.069-0.178 (P < 0.05).
The DL model based on PCP CT images performed acceptably in differentiating adrenal metastases from benign lesions, and it may assist radiologists in accurate tumor staging for patients with a history of extra-adrenal malignancy.
开发基于多期 CT 的深度学习（DL）模型，以区分肾上腺转移瘤与良性病变。
本回顾性的双中心研究纳入了 2008 年 1 月至 2021 年 3 月期间接受多期 CT 扫描的 1059 例患者的 1146 个肾上腺病灶。该研究包括 564 例经手术证实的腺瘤，以及 135 例良性病变和 447 例经观察证实的转移瘤。基于多期 CT 图像开发、验证和测试了 DL 模型。使用准确性（ACC）和受试者工作特征（ROC）曲线比较分类模型、成像阶段和有或没有 DL 的放射科医生的诊断性能。整合判别改善（IDI）分析和 DeLong 检验用于比较模型之间的曲线下面积（AUC）。决策曲线分析（DCA）用于评估预测模型的临床实用性。
基于 LASSO 的 DL 特征（DLSL）的 AUC 高于其他分类模型（IDI>0，P<0.05）。此外，基于平扫期（PCP）的 DLSL 在独立外部验证（AUC=0.881，ACC=82.9%）和临床测试队列（AUC=0.790，ACC=70.4%）中表现最佳，优于基于其他单期或三期图像的 DLSL（IDI>0，P<0.05）。DCA 表明，基于 PCP 的 DLSL 提供了更高的净收益（0.01-0.95）。当放射科医生结合使用 DL 模型时，诊断性能会有显著提高，AUC 提高了 0.056-0.159，ACC 提高了 0.069-0.178（P<0.05）。
基于 PCP CT 图像的 DL 模型在区分肾上腺转移瘤与良性病变方面表现尚可，它可能有助于放射科医生对有肾上腺外恶性肿瘤病史的患者进行准确的肿瘤分期。
Zotero 插件