| 邱宇轩,韩志江.CT平扫最小CT值列线图模型对肾上腺乏脂腺瘤的预测价值[J].浙江中西医结合杂志,2024,34(5): |
| CT平扫最小CT值列线图模型对肾上腺乏脂腺瘤的预测价值 |
| The predictive value of non-enhanced CT minimum attenuation value nomogram model for adrenal lipid-poor adenomas |
| 投稿时间：2023-10-09 修订日期：2024-01-04 |
| DOI： |
| 中文关键词: 肾上腺乏脂腺瘤 体层摄影术,X线计算机 最小CT值 列线图 |
| 英文关键词:Lipid-poor adenoma Computer Tomography Minimum attenuation values Nomogram |
| 基金项目: |
|
| 摘要点击次数: 409 |
| 全文下载次数: 2 |
| 中文摘要: |
| 目的：探讨基于CT平扫最小CT值（minimum attenuation values, minAVs）的列线图模型在肾上腺乏脂腺瘤中的预测价值。
材料与方法: 回顾性收集经病理证实的肾上腺乏脂腺瘤患者的临床、CT影像及病理数据，并与非腺瘤对照，以7:3随机拆分为训练和测试集，采用4点法测量瘤体minAVs、minAVs≤0HU的个数、平均CT值（mean attention values, meaAVs）和直径等。训练集中使用逻辑回归筛选变量及构建列线图模型，测试集验证其诊断效能。受试者操作曲线（Receiver operating characteristic curve, ROC）评估模型诊断效能，Hosmer-Lemeshow检验评估模型拟合优度。
结果:训练集和测试集中乏脂腺瘤分别为73枚、39枚，非腺瘤为71枚、23枚。单因素逻辑回归分析显示，直径、meaAVs、minAVs和minAVs≤0HU的OR（odds ratio）和95%CI （confidence interval）分别为0.95（0.92-0.98）、0.89（0.84-0.94）、0.92（0.89-0.95）、4.50（2.58-7.86）p值均＜0.05，余患者一般资料无统计学差异（p值均＞0.05）。多因素分析中仅直径和minAVs≤0HU存在统计学差异，p值均＜0.05，OR值为0.96（0.93-0.99）、4.92（2.22-10.91）。列线图模型ROC分析中训练集、测试集曲线下面积及95%CI分别为0.855 （0.793-0.917）、0.832（0.731-0.932），敏感度为75.3%、71.8%，特异度为88.7%、87.0%，准确度为81.2%、77.4%，Hosmer-Lemeshow检验p值分别为0.97、0.94。
结论: 基于minAVs≤0HU个数和直径的列线图模型可早期预测肾上腺乏脂腺瘤，为患者个性化诊疗提供依据。 |
| 英文摘要: |
| Purpose: Exploring the predictive value of nomogram models based on non-enhanced CT minimum attenuation values (minAVs) in adrenal lipid-poor adenomas.
Materials and Methods: Retrospectively collected clinical, CT imaging, and pathological data of patients with lipid-poor adenomas confirmed by pathology. They were compared with non-adenomas and randomly divided into training and testing sets in a 7:3 ratio. Four-point method was used to measure the lesion minAVs, the number of minAVs≦0HU, mean attention values (meaAVs), and diameter. Training set is used to select variables and construct a nomogram model using logistic regression, and the test set is used to validate its diagnostic performance. The diagnostic efficacy of the model is evaluated using the Receiver operating characteristic curve (ROC), and the goodness-of-fit of the model is assessed using the Hosmer-Lemeshow test.
Results: The training set had 73 adrenal lipid-poor adenomas and 71 non-adenomas, while the testing set had 39 lipid-poor adenomas and 23 non-adenomas. Logistic regression analysis of univariate factors revealed that the OR (odds ratio) and 95% confidence interval for diameter, meaAVs, minAVs, and minAVs≦0HU were 0.95 (0.92-0.98), 0.89 (0.84-0.94), 0.92 (0.89-0.95), and 4.50 (2.58-7.86), respectively, with all p values < 0.05. Remaining patient baseline data showed no statistically significant differences (p values all > 0.05). In the multivariate analysis, statistical differences were found only between diameter and minAVs≦0HU with both p-values < 0.05. The OR values were 0.96 (0.93-0.99) and 4.92 (2.22-10.91). ROC analysis of the nomogram model revealed that the area under the curve for both the training set and test set was 0.855 (0.793-0.917) and 0.832 (0.731-0.932), respectively. The sensitivity was 75.3% and 71.8%, while the specificity was 88.7% and 87.0%. The accuracy was 81.2% and 77.4%, respectively. The Hosmer-Lemeshow p-values were 0.97 and 0.94, respectively.
Conclusion: The nomogram model with minAVs≦0HU and diameter can prediction adrenal lipid-poor adenoma, offering personalized diagnosis and treatment options for patients. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
