杨晓东,刘屹,郭妍,牛猛,徐克.CT影像组学鉴别最大径≥ 1 cm良恶性胆囊息肉[J].中国医学影像技术,2019,35(12):1842~1846 |
CT影像组学鉴别最大径≥ 1 cm良恶性胆囊息肉 |
Differential diagnosis of benign and malignant gallbladder polypoid lesions with maximum diameter ≥ 1 cm based on CT radiomics |
投稿时间:2019-07-24 修订日期:2019-08-30 |
DOI:10.13929/j.1003-3289.201907174 |
中文关键词: 体层摄影术,X线计算机 磁共振成像 胆囊息肉样变 影像组学 |
英文关键词:tomography, X-ray computed magnetic resonance imaging polypoid lesion of gallbladder radiomics |
基金项目: |
|
摘要点击次数: 1949 |
全文下载次数: 933 |
中文摘要: |
目的 探讨基于CT影像组学鉴别诊断良恶性胆囊息肉样病变(PLG)的价值。方法 145例PLG接受腹部增强CT检查、最大径≥ 1 cm且经病理证实,良性82例,恶性63例,将其随机分为训练集和测试集。使用ITK-SNAP软件手动分割门静脉期CT图像的3D ROI,导入AK分析软件提取影像组学特征,以Lasso回归法进行特征降维,借助R语言软件建立Logistic回归模型并验证,采用ROC曲线评价该模型的诊断效能。结果 最终提取7个与鉴别良恶性PLG相关的特征,基于训练集所得最佳诊断阈值为0.370,模型在测试集中进行验证,发现阈值为0.370时准确率为0.886,特异度和敏感度分别为0.880和0.895,AUC值为0.924。结论 CT影像组学可有效鉴别最大径≥ 1 cm的良恶性PLG。 |
英文摘要: |
Objective To explore the value of radiomics based on CT in differential diagnosis of benign and malignant polypoid lesions of the gallbladder (PLG). Methods Totally 145 patients with PLG ≥ 1 cm who underwent abdominal enhanced CT examination and confirmed by pathology were collected. Among them, benign PLG was found in 82 cases, while malignant ones were detected in 63 cases. The patients were randomly divided into training set and test set. 3D ROIs of portal vein phase CT images were manually segmented using ITK-SNAP software. AK software was introduced to extract high-dimensional radiomics features. Then Lasso regression was used to reduce the dimension of the features. Logistic regression model was established and tested with R language software. Finally, the diagnostic performance of the model was evaluated with ROC. Results Seven features related to benign and malignant identification of gallbladder polyps were obtained. The optimal threshold based on training set was 0.370. After the model was established, the threshold was used for test set with accuracy was 0.886, the specificity and sensitivity was 0.880 and 0.895, respectively, and AUC was 0.924. Conclusion CT radiomics can effectively identify benign and malignant PLG with the maximum diameter ≥ 1 cm. |
查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|