一种用于区分恶性与非恶性胸腔积液的新型临床影像组学标准化方法。
Novel clinical radiomic nomogram method for differentiating malignant from non-malignant pleural effusions.
发表日期：2023 Jul
作者：
Rui Han, Ling Huang, Sijing Zhou, Jiran Shen, Pulin Li, Min Li, Xingwang Wu, Ran Wang
来源：
Disease Models & Mechanisms
摘要：
建立一种临床放射组学评分模型，以区分恶性和非恶性胸腔积液。本研究纳入了146例恶性胸腔积液（MPE）患者和93例非恶性胸腔积液（NMPE）患者。使用CT提取胸部病变的感兴趣区域图像特征。进行单变量分析，并利用最小绝对收缩选择算子和多变量logistic回归分析筛选放射组学特征并计算放射组学评分。将临床因素与放射组学评分相结合，构建了一种评分卡。采用ROC曲线和决策曲线分析（DCA）评估预测效果。经过筛选，选择了19个放射组学特征和2个临床因素作为最佳预测因子，建立了结合模型，构建了一种评分卡。训练队列中，结合模型的AUC为0.968（95%置信区间[CI] = 0.944-0.986），验证队列中为0.873（95% CI = 0.796-0.940）。结合模型的AUC值显著高于临床和放射组学模型的AUC值（分别为0.968 vs. 0.874和0.878）。验证队列中的结果与之类似（分别为0.873, 0.764和0.808）。DCA证实了放射组学评分卡的临床实用性。基于CT的放射组学显示出比临床和放射学特征更好的诊断准确性和模型拟合度，可区分MPE和NMPE。结合两者可以获得更好的诊断表现。这些发现支持使用胸部CT诊断MPE的评分卡在临床应用中的使用。© 2023 Published by Elsevier Ltd.
To establish a clinical radiomics nomogram that differentiates malignant and non-malignant pleural effusions.A total of 146 patients with malignant pleural effusion (MPE) and 93 patients with non-MPE (NMPE) were included. The ROI image features of chest lesions were extracted using CT. Univariate analysis was performed, and least absolute shrinkage selection operator and multivariate logistic analysis were used to screen radiomics features and calculate the radiomics score. A nomogram was constructed by combining clinical factors with radiomics scores. ROC curve and decision curve analysis (DCA) were used to evaluate the prediction effect.After screening, 19 radiomics features and 2 clinical factors were selected as optimal predictors to establish a combined model and construct a nomogram. The AUC of the combined model was 0.968 (95% confidence interval [CI] = 0.944-0.986) in the training cohort and 0.873 (95% CI = 0.796-0.940) in the validation cohort. The AUC value of the combined model was significantly higher than those of the clinical and radiomics models (0.968 vs. 0.874 vs. 0.878, respectively). This was similar in the validation cohort (0.873, 0.764, and 0.808, respectively). DCA confirmed the clinical utility of the radiomics nomogram.CT-based radiomics showed better diagnostic accuracy and model fit than clinical and radiological features in distinguishing MPE from NMPE. The combination of both achieved better diagnostic performance. These findings support the clinical application of the nomogram in diagnosing MPE using chest CT.© 2023 Published by Elsevier Ltd.