首页|IVIM-DWI直方图参数鉴别良恶性甲状腺结节

IVIM-DWI直方图参数鉴别良恶性甲状腺结节

Diagnostic of intravoxel incoherent motion diffusion-weighted imaging histogram parameters in distinguis-hing between benign and malignant thyroid nodules

扫码查看
目的:探讨体内不相干运动(IVIM)DWI直方图参数对甲状腺良、恶性结节的鉴别诊断效能.方法:回顾性纳入2017年3月—2022年9月在本院经手术病理证实的51例甲状腺结节患者的病例资料,其中良性组24例、恶性组27例.每例患者术前行多b值小视野DWI检查,经图像后处理生成真扩散系数(D)、伪扩散系数(D*)和灌注分数(f)伪彩图及b值为990 s/mm2的表观扩散系数(ADC)图.手动逐层勾画整个病灶的ROI并生成容积感兴趣区(VOI).利用matlab自编程序,分别在D图、D*图与f图上测量得到每个肿瘤VOI的第5、15、85和95百分位数(P5,P15,P85,P95)值、均值、四分位距、偏度、均方根差和变异系数,同时计算肿瘤VOI处的ADC均值.使用t检验或Mann-Whitney U检验比较肿瘤的IVIM-DWI直方图参数、ADC均值和临床资料的组间差异,将有统计学意义(P<0.05)的变量采用逻辑回归方法建立联合模型.采用受试者工作特征(ROC)曲线分析联合模型鉴别甲状腺良恶性结节的诊断效能.结果:患者性别、平均ADC值,D图的P5、P15、P85和P95值、均值、偏度和均方根差,以及D*图的P5和P15值、均值、偏度和变异系数,在良、恶性组之间的差异均具有统计学意义(P<0.05).在有统计学意义的直方图参数中最终筛选出3个最佳参数,即D图的偏度、P15值和均方根差(P均<0.005).平均ADC值、性别联合D图的偏度、P15值和均方根差构建的组学联合模型的AUC为0.94(95%CI:0.84~0.99),敏感度为88.46%,特异度为90.91%;平均ADC值联合性别构建的常规联合模型的AUC为0.86(95%CI:0.74~0.94),敏感度为77.78%,特异度为87.50%.结论:与常规ADC值联合性别构建的模型相比,IVIM-DWI直方图参数联合ADC值和性别构建的联合模型可提高甲状腺良恶性结节的诊断效能.
Objective:The purpose of this study was to investigate the diagnostic efficacy of his-togram parameters extracted from intravoxel incoherent motion(IVIM)DWI in distinguishing be-tween benign and malignant thyroid nodules.Methods:The clinical data of 51 patients with thyroid nodule disease confirmed by surgery and pathology from March 2017 to September 2022 were retro-spectively included,including 24 cases of benign nodules and 27 cases of malignant nodules,all con-firmed by surgical pathology.Preoperative multi-b-value small field-of-view diffusion-weighted imaging was performed to generate maps of the true diffusion coefficient(D),the pseudo-diffusion coefficient(D*),and the perfusion fraction(f),as well as the apparent diffusion coefficient(ADC)map with b value of 990s/mm2.The entire lesion volume of the region of interest(VOI)was manually delineated.Using a custom-made MATLAB program,the 5-,15-,85-,and 95-percentile(named as P5,P15,P85 and P95),mean,interquartile range,skewness,root-mean-square deviation,and coefficient of variation were calculated for the D-,D*-,and f-maps within the VOI,as well as the mean ADC value within the VOI.The intergroup differences of IVIM histogram parameters,average ADC values and basic clinical information were compared using t-test or Mann-Whitney U-test,and then the statistically significant variables with P<0.05 were selected out for establishing a combined model using logistic regression method.The diagnostic efficacy of the combined model in distinguishing between benign and malignant thyroid nodules was then evaluated using receiver operating characteristic(ROC)curve analysis.Re-sults:There were significant intergroup differences in gender,the values of mean ADC,P5,P15,P85 and P95,skewness,and root-mean-square deviation of the nodules on the D-map,and P5,P15,mean,skewness,and coefficient of variation on the D*-map(all P<0.05).Three optimal histogram parame-ters were finally selected(P<0.005),including skewness,P15 value and root-mean-square deviation on the D-map.The AUC of the combined radiomics model constructed by mean ADC value,gender combined with skewness,P15 value,and root-mean-square deviation on the D-map was 0.94(95%CI:0.84~0.99),with a sensitivity of 88.46%and a specificity of 90.91%.The AUC of the conventional combined model constructed by mean ADC value and gender was 0.86(95%CI:0.74~0.94),with a sensitivity of 77.78%and a specificity of 87.50%.Conclusion:Compared with the prediction model con-structed by average ADC value and gender,the combined model constructed with IVIM-DWI histo-gram parameters,ADC value and gender can improve the diagnostic efficacy for benign and malignant thyroid nodules.

Thyroid nodulesIntravoxel incoherent motionDiffusion-weighted imagingHis-togramDiagnostic model

任杰、李兴鹏、周荣杰、冯慧杰、沈智威、岳云龙

展开 >

100038 北京,首都医科大学附属北京世纪坛医院磁共振室

100038 北京,首都医科大学附属北京世纪坛医院放射科

100600 北京,飞利浦健康医疗有限公司

甲状腺结节 体素内不相干运动 扩散加权成像 直方图 诊断模型

2024

放射学实践
华中科技大学同济医学院

放射学实践

CSTPCD北大核心
影响因子:1.08
ISSN:1000-0313
年,卷(期):2024.39(12)