基于分数阶微积分模型DWI全容积直方图分析对乳腺良恶性病变的鉴别诊断价值  

The value of whole-volume histogram analysis of DWI based on fractional-order calculus model in differential diagnosis of benign and malignant breast lesions

作　　者：张宝媞 汪飞 孙宜楠 章少兰 汪敏敏 高何燕 程兰兰[2] 刘孟潇 朱娟[1,2] ZHANG Bao-ti;WANG Fei;SUN Yi-nan(Department of Medical Imaging,Anqing Medical Center,the Fifth Clinical College of Anhui Medical University,246003 Anhui,China)

机构地区：[1]安徽医科大学第五临床医学院(安庆医学中心)医学影像科,安徽安庆246003 [2]安庆市立医院医学影像科,安徽安庆246003 [3]西门子医疗系统有限公司MR科研市场部,上海200126

出　　处：《放射学实践》2025年第3期331-338,共8页Radiologic Practice

基　　金：安徽省教育厅高校科学研究重点项目(2023AH053433)。

摘　　要：目的:探讨基于分数阶微积分(FROC)模型DWI全容积直方图参数在鉴别乳腺良、恶性病变中的应用价值。方法:回顾性分析本院91例乳腺良恶性病变患者的临床及影像资料。所有患者均行常规MRI平扫、动态增强扫描及多b值DWI扫描,经后处理得到FROC-DWI[异常扩散系数(D)、体素内扩散异质性参数(β)和空间参数(μ)]及ADC图全容积直方图参数。采用独立样本t检验或Mann-Whitney U检验比较良恶性病变组的D、β、μ及ADC图全容积直方图参数的差异。通过二元Logistic回归构建综合模型,采用受试者操作特征(ROC)曲线评价综合模型及各直方图参数对乳腺良、恶性病变的诊断效能。结果:除D_(熵)、β_(最大值、峰度、90th、最小值)及ADC_(熵)外,其他FROC-DWI及ADC图全容积直方图参数在乳腺良、恶性病变组间差异均有统计学意义(P<0.05)。在各直方图参数组中,D50th、β_熵、μ偏度、ADC50th的ROC曲线下面积(AUC)最高(分别为0.889、0.813、0.759、0.894)。纳入组间差异有统计学意义的FROC-DWI及ADC图全容积直方图参数建立综合模型1,AUC为0.960;纳入组间差异有统计学意义的FROC-DWI全容积直方图参数建立综合模型2,AUC为0.962。结论:基于FROC-DWI全容积直方图参数在鉴别乳腺良、恶性病变中具有一定价值,在各直方图参数组中,D50th、β_(熵)、μ_(峰度)、ADC_(50th)的诊断价值最高;多直方图参数联合诊断可进一步提高诊断效能。Objective:The purpose of this study was to investigate the application value of whole-volume histogram parameters of DWI based on fractional-order calculus(FROC)model for differentiating benign and malignant breast lesions.Methods:The clinical and imaging data of 91 patients with benign and malignant breast lesions in our hospital were retrospectively analyzed.All patients underwent conventional MRI plain scanning,dynamic enhanced scanning and multi-b-value DWI scanning.Full volume histogram parameters for FROC-DWI[the diffusion coefficient D,fractional-order derivative in spaceβ,and a spatial parameterμ]and ADC maps were obtained by post-processing.Independent samples t-test or Mann-Whitney U-test was used to compare the differences in whole-volume histogram parameters of D、β、μand ADC maps in benign and malignant lesion groups.Comprehensive models were constructed by binary logistic regression,and the diagnostic efficacy of the comprehensive models and each histogram parameter for benign and malignant breast lesions were evaluated by using receiver operating characteristic(ROC)curve.Results:All whole-volume histogram parameters of FROC-DWI and ADC maps except D entropy,β_(max,Kurtosis,μ90th,min)and ADC_(entropy)exhibited significant differences between the two groups(P<0.05).D 50th,βentropy,μskewness,ADC 50th had the highest area under curve(AUC)(0.889,0.813,0.759 and 0.894,respectively)in their individual histogram parameter groups.Inclusion of whole-volume histogram parameters of FROC-DWI and ADC maps with statistically significant differences between groups to build a comprehensive model 1,and AUC was 0.960;inclusion of whole-volume histogram parameters of FROC-DWI with statistically significant differences between groups to build a comprehensive model 2,and an AUC of 0.962.Conclusion:The whole-volume histogram parameters of FROC-DWI have some value in identifying benign and malignant breast lesions,and D_(50th),β_(entropy,μkurtosis)and ADC_(50th)have the highest diagnostic value among histogram

关 键 词：乳腺肿瘤 分数阶微积分模型 直方图 扩散加权成像 磁共振成像 

分 类 号：R445.2[医药卫生—影像医学与核医学] R737.9[医药卫生—诊断学]