基于超声影像组学诺模图预测甲状腺TI-RADS 4〜5类结节良恶性  

A Nomogram Based on Ultrasound Radiomics Model for Discrimination Between Malignant and Benign in TI-RADS 4-5 Thyroid Nodules

作　　者：石艳萍 蒋艳 王嘉俊 孙亚 徐井旭 郭君[1] 梁蕾[1] Shi Yanping;Jiang Yan;Wang Jiajun;Sun Ya;Xu Jingxu;Guo Jun;Liang Lei(Department of Ultrasound,Aerospace Center Hospital,Beijing 100049,China;Department of Research Collaboration,R&D Center,Beijing Dccpwisc&League of PHD Technology Co.,Ltd.,Beijing 100080,China)

机构地区：[1]航天中心医院超声科,北京市100019 [2]北京深睿博联科技有限责任公司研究协作部,研发部,北京市100080

出　　处：《中国超声医学杂志》2021年第12期1328-1332,共5页Chinese Journal of Ultrasound in Medicine

基　　金：海淀培育项目(No.HP2021-32-50702);2019年航天中心医院科研基金项目(No.YN201910)。

摘　　要：目的通过超声影像组学联合临床因素建立模型,评估其对甲状腺影像报告和数据系统(TI-RADS) 4~5类甲状腺结节的诊断价值。方法回顾性研究共纳入208例患者。收集患者相关临床危险因素建立临床模型。所有患者在穿刺或术前接受常规超声检查。将图像导入平台后人工描绘结节感兴趣区,以病理结果为参照进行机器学习,然后提取并筛选组学特征建立超声影像组学模型。最终临床与影像组学结合建立联合模型。通过受试者工作特征曲线和决策曲线比较各模型的诊断效能和临床净效益。结果经验证,临床-影像组学联合模型(AUC:0.85)比超声影像组学模型(AUC:0.81)和临床模型(AUC:0.71)具有更高的预测价值。决策曲线分析还表明,联合模型在较为广泛的高风险阈值范围内比超声影像组学模型或临床模型具有更高的净效益。结论与单纯评价甲状腺癌相关的临床危险因素或超声影像组学评分相比,临床-影像组学联合模型在诊断效能和临床净效益方面都能提高TI-RADS 4~5类甲状腺结节的预测能力。Objective To evaluate the diagnostic value of a clinical-based model, a ultrasound-based radiomics model, and a clinical-radiomics combined model for predicting thyroid imaging reporting and data system(TI-RADS) 4-5 Thyroid Nodules. Methods A total of 208 patients were included in this retrospective study. Clinical risk factors were collected to establish clinical model. All patients received routine ultrasound examination before puncture or operation. After the images were imported into the platform, the region of interest(ROI) of the nodules were manually depicted, and then machine learning was performed with the pathological results as reference. Then, the radiomics features were extracted and screened to establish the ultrasound radiomics model. Finally, the combined model was established by the combination of clinical and ultrasound radiomics model. The diagnostic efficacy and clinical net benefit of each model were compared by receiver operating characteristic curve and decision curve. Results The combined model(area under curve, AUC: 0.85) had higher predictive value than ultrasound radiomics model(AUC: 0.81) and clinical model(AUC: 0.71). The decision curve analysis also showed that the combined model had higher net benefits in a wide range of high risk threshold than either the radiomics model or the clinical model. Conclusions The combined model can improve the accuracy of TI-RADS 4-5 thyroid nodules predictions both in terms of diagnostic performance and clinical net benefit, compared with evaluating only clinical risk factors or radiomics score associated with thyroid cancer.

关 键 词：甲状腺结节 影像组学 联合模型 诺模图 临床危险因素 

分 类 号：R445.1[医药卫生—影像医学与核医学] R736.1[医药卫生—诊断学]