| 张云姣,刘志聪,蔡 洁.超声影像特征用于乳腺良恶性肿瘤鉴别诊断的Logistic回归分析及风险预测模型建立[J].肿瘤学杂志,2016,22(3):214-217. |
| 超声影像特征用于乳腺良恶性肿瘤鉴别诊断的Logistic回归分析及风险预测模型建立 |
| Study on Logistic Regression Analysis and Risk Prediction Model Establishment with Ultrasound Image Features for Differential Diagnosis of Breast Neoplasms |
| 投稿时间：2015-08-27 |
| DOI：10.11735/j.issn.1671-170X.2016.03.B012 |
|
 |
| 中文关键词: 乳腺肿瘤 超声检查 Logistic回归模型 鉴别诊断 |
| 英文关键词:breast neoplasm ultrasonography Logistic regression model differential diagnosis |
| 基金项目: |
|
| 摘要点击次数: 2030 |
| 全文下载次数: 864 |
| 中文摘要: |
| 摘 要：［目的］ 建立基于超声影像特征预测乳腺良恶性肿瘤的Logistic回归模型，并探讨预测模型在鉴别乳腺良恶性肿瘤中的应用价值。［方法］ 回顾性收集经病理证实的288例乳腺肿瘤患者，其中良性肿瘤、恶性肿瘤各144例，比较两组彩色多普勒超声各项指标特征，以病理诊断结果作为金标准，建立Logistic回归模型，绘制ROC曲线并计算曲线下面积。［结果］多因素Logistic回归分析结果显示：乳腺肿瘤形态(OR=7.149)、肿瘤边界(OR=8.908)、回声均匀度(OR=3.374)、包膜情况(OR=13.079)、蟹足或毛刺(OR=15.690)是乳腺良恶性肿瘤鉴别诊断的主要超声影像特征指标。Logistic回归模型对乳腺良恶性肿瘤的预测准确性为94.4%(272/288)，敏感性为91.7%(132/144)，特异性为97.2%(140/144)，阳性预测价值97.1%(132/136)，阴性预测价值92.1%(140/152)。ROC曲线下面积为0.944±0.016（P<0.001，95%CI：0.914~0.975）。［结论］ 基于超声影像特征的Logistic预测模型对于鉴别乳腺良恶性肿瘤具有较高的价值，可用于指导临床实践。 |
| 英文摘要: |
| Abstract：［Purpose］ To establish the Logistic regression prediction model on ultrasound image feature in differential diagnosis of breast neoplasm，and to evaluate the diagnostic value of Logistic model in differentiating malignant and benign breast tumors. ［Methods］ The ultrasound image features of 288 cases with breast neoplasms were divided into benign tumor group and malignant tumor 144 cases in each group. A Logistic model was obtained on the basis of uhrasonographic features，receiver operator charaeters(ROC) curve was constructed to assess the performance of the Logistic model. ［Results］ Multiariate Logistic regression analysis showed that breast tumor morphology(OR=7.149)，tumor boundary(OR=8.908)，echo homogeneity(OR=3.374)，envelope(OR=13.079)，crab foot or burr(OR=15.690) were the main features in the differential diagnosis of benign and malignant breast tumors. The accuracy of Logistic regression model for breast cancer was 94.4%(272/288)，sensitivity was 91.7%(132/144)，specificity was 97.2%(140/144)，positive predictive value was 97.1%(132/136)，and negative predictive value was 92.1%(140/152). The area under the ROC curve was 0.944±0.016（P<0.001，95%CI：0.914~0.975）. ［Conclusion］ Logistic regression prediction model of ultrasound image feature plays an important role in the differential diagnosis of breast neoplasm. |
|
在线阅读
查看全文 查看/发表评论 下载PDF阅读器 |