| 牛雅宁,于一行,龚毓宾,董健,赵婧,胡文笳,董长宪,刘秋雨,吴刚.灰阶超声影像组学鉴别诊断皮下组织血管瘤与卡波西型血管内皮瘤[J].中国医学影像技术,2022,38(11):1704~1709 |
| 灰阶超声影像组学鉴别诊断皮下组织血管瘤与卡波西型血管内皮瘤 |
| Gray-scale ultrasound-based radiomics for differential diagnosis of subcutaneous hemangioma and Kaposiform hemangioendothelioma |
| 投稿时间：2022-06-14 修订日期：2022-08-25 |
| DOI：10.13929/j.issn.1003-3289.2022.11.024 |
| 中文关键词: 血管瘤 血管内皮瘤 超声检查 影像组学 |
| 英文关键词:hemangioma hemangioendothelioma ultrasonography radiomics |
| 基金项目: |
|
| 摘要点击次数: 825 |
| 全文下载次数: 370 |
| 中文摘要: |
| 目的 观察灰阶超声影像组学鉴别诊断皮下组织血管瘤（HE）与卡波西型血管内皮瘤（KHE）的价值。方法 回顾性分析143例皮下组织HE和70例KHE共252处病灶，按7：3比例将病灶随机分为训练集（n=176）和验证集（n=76）；提取病灶灰阶超声影像组学特征，构建影像组学模型，结合临床资料建立联合模型，观察各模型鉴别诊断皮下组织HE与KHE的效能。结果 共选取22个系数非零的稳定特征。影像组学模型鉴别训练集皮下组织HE与KHE的曲线下面积（AUC）、准确率、敏感度、特异度、阳性预测值及阴性预测值分别为0.91[95%CI（0.89，0.93）]、91.41%、83.20%、93.92%、95.79%及89.00%；用于验证集分别为0.85[95%CI（0.83，0.87）]、90.78%、79.32%、97.90%、96.71%及88.68%。联合模型鉴别训练集皮下组织HE与KHE的AUC、准确率、敏感度、特异度、阳性预测值及阴性预测值分别为0.94[95%CI（0.92，0.96）]、94.33%、90.77%、96.38%、94.23%及94.90%；用于验证集分别为0.90[95%CI（0.88，0.92）]、92.14%、85.69%、95.76%、93.33%及92.30%。联合模型鉴别诊断皮下组织HE与KHE的AUC均大于影像组学模型（P均<0.05）。结论 灰阶超声影像组学鉴别诊断皮下组织HE与KHE的效能较佳；联合临床特征可进一步提高其诊断效能。 |
| 英文摘要: |
| Objective To observe the value of gray-scale ultrasound-based radiomics for differential diagnosis of subcutaneous hemangioma (HE) and Kaposiform hemangioendothelioma (KHE).Methods Data of 252 lesions of 143 children with subcutaneous HE and 70 children with KHE were retrospectively analyzed. The lesions were randomly divided into training set (n=176) and validation set (n=76) at the ratio of 7:3. Gray-scale ultrasound-based radiomics features were extracted to build radiomics models and combined model with clinical data. The efficacy of each model for differential diagnosis of subcutaneous HE and KHE were observed.Results A total of 22 stable features with non-zero coefficients were selected. The area under the curve (AUC), accuracy, sensitivity, specificity, positive predictive value and negative predictive value of radiomics model for differential diagnosis of subcutaneous HE and KHE in training set was 0.91 (95%CI[0.89, 0.93]), 91.41%, 83.20%, 93.92%, 95.79% and 89.00%, while in validation set was 0.85 (95%CI[0.83, 0.87]), 90.78%, 79.32%, 97.90%, 96.71% and 88.68%, respectively. The AUC, accuracy, sensitivity, specificity, positive predictive value and negative predictive value of combined model for differential diagnosis of subcutaneous HE and KHE in training set was 0.94 (95%CI[0.92, 0.96]), 94.33%, 90.77%, 96.38%, 94.23% and 94.90%, while in validation set was 0.90 (95%CI[0.88, 0.92]), 92.14%, 85.69%, 95.76%, 93.33% and 92.30%, respectively. The AUC of combined model for differential diagnosis of subcutaneous HE and KHE were higher than those of radiomics model (both P<0.05).Conclusion Gray-scale ultrasound-based radiomics was effective for differentiating subcutaneous HE and KHE, and the diagnostic efficacy could be further improved combining with clinical features. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
