实用医学杂志 ›› 2025, Vol. 41 ›› Issue (21): 3442-3448.doi: 10.3969/j.issn.1006-5725.2025.21.021

• 医学检查与临床诊断 • 上一篇    

常规影像和CT影像组学特征鉴别腮腺腺淋巴瘤与恶性肿瘤的价值

文国良1,方行1,张卫2,3()   

  1. 1.桂林医学院附属医院放射科 (广西 桂林 541001 )
    2.柳州市人民医院放射科 (广西 柳州 545006 )
    3.柳州市分子影像重点实验室 (广西 柳州 545006 )
  • 收稿日期:2025-07-30 出版日期:2025-11-10 发布日期:2025-11-13
  • 通讯作者: 张卫 E-mail:holly2yang@126.com
  • 基金资助:
    柳州市人民医院项目(LRYGCC202204)

Value of conventional radiological features and ct radiomics features in differentiating parotid adenolymphoma from malignant tumors

Guoliang WEN1,Hang FANG1,Wei. ZHANG2,3()   

  1. *.Department of Radiology,Affiliated Hospital of Guilin Medical University,Guilin 541001,Guangxi,China
  • Received:2025-07-30 Online:2025-11-10 Published:2025-11-13
  • Contact: Wei. ZHANG E-mail:holly2yang@126.com

摘要:

目的 探讨常规影像和CT影像组学特征区分腮腺腺淋巴瘤与恶性肿瘤的价值。 方法 回顾性收集128例腺淋巴瘤和39例腮腺恶性肿瘤患者的影像资料。分析影像组学特征获得影像组学评分,并构建影像组学模型。分析常规影像得到独立危险因素,并构建常规影像模型。最后影像组学评分联合常规影像建立综合模型并绘制列线图。采用Delong检验两两比较模型间的差异。 结果 基于CT静脉期图像筛选出9个最有价值的影像组学特征,常规影像经单因素及多因素logistic回归分析发现形状、强化程度是腮腺腺淋巴瘤与恶性肿瘤的独立预测因素。影像组学模型优于常规影像模型(P < 0.05),影像组学模型的曲线下面积(AUC)为0.938(95%CI:0.887 ~ 0.988),特异度、灵敏度分别为0.856、0.923。常规影像、影像组学评分经单因素及多因素logistic回归分析发现影像组学评分、强化程度是腮腺腺淋巴瘤与恶性肿瘤的独立危险因素,影像组学评分联合强化程度建立的综合模型诊断效能与影像组学模型的诊断效能差异无统计学意义(P > 0.05)。 结论 CT影像组学模型鉴别腮腺腺淋巴瘤与恶性肿瘤的性能优于常规影像模型。综合模型诊断效能得到提高,有助于临床决策。

关键词: 腮腺, 腺淋巴瘤, 恶性肿瘤, 影像组学

Abstract:

Objective To investigate the value of conventional radiological features and CT radiomics features to differentiate parotid adenolymphoma from malignant tumor. Methods Radiological data from 128 patients with adenolymphoma and 39 patients with parotid malignancy were collected between December 2018 and October 2023. Radiomics features were extracted to obtain Rad?score and construct a radiomics model. Conventional radiological features were analyzed to obtain independent predictors, and a conventional radiological model was constructed. Model performance was compared by the DeLong test, and the Rad?score was combined with radiological features to establish a comprehensive model and plot a nomogram. Results Based on CT venous phase images, 9 optimal radiomic features were selected. Conventional radiological features were analyzed by univariate and multivariate logistic regression analysis and found that shape and degree of enhancement were independent predictors of adenolymphoma and malignant parotid tumors. The CT radiomics model is superior to the conventional radiological model(P < 0.05). The area under the curve (AUC) of the radiomics model was 0.938 (95% CI: 0.887 ~ 0.988), and the specificity and sensitivity were 0.856 and 0.923, respectively. Conventional radiological features and Rad?score were analyzed by univariate and multivariate logistic regression analysis and found that Rad?score and degree of enhancement were independent predictors of adenolymphoma and malignant parotid tumors. The difference between the diagnostic efficacy of the comprehensive model created by the Rad?score combined with the degree of enhancement and the diagnostic efficacy of the radiomics model was not statistically significant (P > 0.05). Conclusion The CT radiomics model performed better than the conventional radiological model in discriminating parotid gland adenolymphoma from malignant tumors. The diagnostic performance of the comprehensive model has improved, aiding in clinical decision?making.

Key words: parotid gland, adenolymphoma, malignant tumor, radiomics

中图分类号: