基于超声特征的不同年龄段人群腮腺肿瘤恶性风险预测模型的建立  

Development of Ultrasound-based Malignancy Risk Predictive Model of Parotid Neoplasms in Different Age Groups

作　　者：吕宗烨 阮骊韬[1] Lyu Zongye;Ruan Litao(Department of Ultrasound Medicine,The First Affiliated Hospital of Xi'an Jiaotong University,Xi'an 71oo6l,China)

机构地区：[1]西安交通大学第一附属医院超声医学科,西安市710061

出　　处：《中国超声医学杂志》2023年第6期618-622,共5页Chinese Journal of Ultrasound in Medicine

摘　　要：目的 基于超声特征构建不同年龄段患者腮腺肿瘤恶性风险预测模型,评估诊断效能。方法 回顾性分析收治并经手术病理证实的320例腮腺肿瘤患者超声资料,依据年龄分为青、中、老年组,通过单因素及多因素分析,得出有统计学意义的腮腺良性、恶性肿瘤鉴别特征,建立预测模型,评价预测效能。结果 青年组边界、形态,中年组形态、后方回声、腮腺及周围淋巴结,老年组边界、形态、腮腺及周围淋巴结是诊断腮腺恶性肿瘤的独立预测因素。青、中、老年组预测模型预测腮腺肿瘤恶性风险曲线下面积分别为0.907、0.872、0.839,预测效能良好。校准曲线表明模型预测的腮腺肿瘤恶性风险与实际发生风险一致性较好。结论 不同年龄段人群腮腺良性、恶性肿瘤超声特征具有差异性,所构建的预测模型可以较为准确地预测腮腺肿瘤恶性风险。Objective To establish and evaluate an ultrasound-based malignancy risk predictive model of parotid neoplasms in different age groups.Methods Retrospectively analyzed ultrasound image data of 320 patients with parotid neoplasms confirmed by surgical pathology,based on age,patients were divided into youth,middle age and elder groups.Through univariate and multivariate analysis,the characteristics of benign and malignant parotid neoplasms were obtained with statistical significance,and the predictive model was established to evaluate the prediction efficiency.Results Boundary and shape in the youth group,shape,posterior acoustic enhancement and abnormal lymph nodes in parotid gland and neck in the middle age group,and boundary,shape and abnormal lymph nodes in parotid gland and neck in the elder group were independent risk factors for parotid cancers.The areas under the curve of predictive models in the three age groups were 0.907,0.872,0.839,which indicated good prediction efficiency.The calibration curve showed good agreement between predicted probability and actual probability of parotid cancers.ConclusionsTThere are significant differences on the sonographic features of benign and malignant parotid neoplasms in different age groups.The models show high accuracy in predicting parotid cancers.

关 键 词：预测模型 腮腺肿瘤 超声 年龄 

分 类 号：R445.1[医药卫生—影像医学与核医学] R739.8[医药卫生—诊断学]