Department of Ultrasound Medicine, The First Affiliated Hospital, Xi'an Jiaotong University, Xi'an 710061, China.
Department of Ultrasound Medicine, The First Affiliated Hospital, Xi'an Jiaotong University, Xi'an 710061, China.
Clin Radiol. 2019 Apr;74(4):306-313. doi: 10.1016/j.crad.2019.01.004. Epub 2019 Feb 10.
To investigate the role of utilizing size-based thresholds of stiffness gradient in diagnosing solid breast lesions and optimizing original Breast Imaging-Reporting And Data System (BI-RADS) classifications.
Two-hundred and twenty-seven consecutive women underwent shear-wave elastography (SWE) before ultrasound-guided biopsy, and 234 solid breast lesions categorized as BI-RADS 3-5 were analysed. Receiver operating characteristic curve analysis was performed based on histopathology. Diagnostic performance among SWE, BI-RADS, and their combination were compared.
The stiffness gradient correlated with the standard deviation of elasticity (SD, r=0.90), and with Tozaki's pattern classification (r=0.64). The area under the receiver operating characteristic curves (AUC) for stiffness gradient (0.939) outperformed SD (0.897) or colour pattern (0.852). Due to significant association with lesion size (r=0.394, p<0.001), stiffness gradient's size-based thresholds (lesions >15 mm: 82.5 kPa; lesions ≤15 mm: 51.1 kPa) were established to reclassify BI-RADS 3-4b lesions. Upgrading category 3 lesions (over the corresponding cut-off value, 3 to 4a) and downgrading categories 4a-4b lesions (less than or equal to the corresponding cut-off value, 4b to 4a, 4a to 3), yielded significant improvement in specificity (90.28% versus 77.78%, p<0.001) and AUC (0.948 versus 0.926, p=0.035) than BI-RADS alone. No significant loss emerged in the sensitivity (88.89% versus 91.11%, p=0.500).
Stiffness gradient exhibited better discriminatory ability than SD or four-colour pattern classification in determining solid breast lesions and applying its size-specific thresholds to categorize BI-RADS 3-4b lesions could improve diagnostic performance.
探讨利用基于硬度梯度的大小阈值诊断实性乳腺病变并优化原始乳腺影像报告和数据系统（BI-RADS）分类的作用。
对 227 例连续行超声引导活检的女性进行剪切波弹性成像（SWE）检查，分析 234 个 BI-RADS 3-5 类实性乳腺病变。基于组织病理学进行受试者工作特征曲线分析。比较 SWE、BI-RADS 及其组合的诊断性能。
硬度梯度与弹性的标准差（SD）（r=0.90）和 Tozaki 模式分类（r=0.64）相关。硬度梯度的受试者工作特征曲线下面积（AUC）（0.939）优于 SD（0.897）或彩色模式（0.852）。由于与病变大小有显著相关性（r=0.394，p<0.001），建立了基于硬度梯度的大小阈值（病变>15mm：82.5kPa；病变≤15mm：51.1kPa）来重新分类 BI-RADS 3-4b 病变。升级类别 3 病变（超过相应的截止值，3 级至 4a 级）和降级类别 4a-4b 病变（小于或等于相应的截止值，4b 级至 4a 级，4a 级至 3 级），特异性（90.28%比 77.78%，p<0.001）和 AUC（0.948 比 0.926，p=0.035）显著提高，而敏感性（88.89%比 91.11%，p=0.500）无显著下降。
硬度梯度在确定实性乳腺病变方面比 SD 或四色模式分类具有更好的鉴别能力，应用其大小特异性阈值对 BI-RADS 3-4b 病变进行分类可提高诊断性能。
