Four non-Gaussian distributed diffusion imaging parameters for differentiating breast imaging reporting and data system MRI category 4 benign and malignant breast tumors
10.13929/j.issn.1003-3289.2025.09.028
- VernacularTitle:4种非高斯分布弥散成像参数鉴别乳腺影像报告和数据系统MRI 4类乳腺良、恶性病变
- Author:
Miaomiao DING
1
;
Zhaoqi LAI
1
;
Yun SU
1
;
Xinyin CHEN
1
;
Xiang ZHANG
1
Author Information
1. 中山大学孙逸仙纪念医院放射科,广东 广州 510120
- Publication Type:Journal Article
- Keywords:
breast neoplasms;
diagnosis,differential;
diffusion magnetic resonance imaging
- From:
Chinese Journal of Medical Imaging Technology
2025;41(9):1586-1590
- CountryChina
- Language:Chinese
-
Abstract:
Objective To explore the efficacy of single and combined parameters from 4 non-Gaussian diffusion models for differentiating breast imaging reporting and data system(BI-RADS)MRI category 4 benign and malignant breast tumors.Methods A total of 161 BI-RADS MRI category 4 breast lesions from 159 patients were retrospectively enrolled.Based on pathological results,the lesions were divided into malignant group(n=132)and benign group(n=29).The apparent diffusion coefficient(ADC)values were calculated from diffusion weighted imaging(DWI)sequences.Multi-b-value diffusion imaging data were acquired and fitted using 4 non-Gaussian models to obtain respective parameters,including diffusion kurtosis imaging(DKI),stretched exponential model(SEM),continuous-time random walk(CTRW)and fractional order calculus(FROC)model.Univariable and multivariable logistic regression analyses were employed to identify the diffusion quantitative indicators useful for differentiating benign and malignant BI-RADS MRI category 4 breast tumors.Receiver operating characteristic(ROC)curves were drawn,and the optimal threshold was determined using Youden index.The differentiating performance of ADC value,single parameter and their combination from non-Gaussian diffusion models were assessed and compared according to the area under the curve(AUC)of ROC curves,as well as the sensitivity,specificity and accuracy under the optimal thresholds.Results In malignant group,αCTRW,KDKI and μFROC values were higher,while ADC,DCTRW,DFROC,DDCSEM,αSEM and DDKI values were lower than those in benign group(all P＜0.05).Multivariable logistic regression analysis identified DCTRW and αCTRW values as independent factors for differentiating benign and malignant BI-RADS MRI category 4 breast tumors(both P＜0.05),and a combined model was then constructed.The AUC of the combined model was higher than that of each single parameter including ADC,DCTRW and αCTRW values(all P＜0.05).Conclusion The combined model of DCTRW and αCTRW had better efficacy than each single parameter for differentiating benign and malignant BI-RADS MRI category 4 breast tumors.
