超高b值扩散加权成像简化协议用于乳腺癌检测
该前瞻性研究纳入127例有不同临床指征行乳腺MRI的女性,评估基于超高b值DWI(b1500和b2500)的非增强简化协议检测和表征可疑乳腺病变的能力,并与增强图像比较。结果显示,b2500简化协议(DWI+ADC+平扫T1/T2)的阴性预测值为99.5%、敏感度97.4%、特异度88.4%,诊断准确性接近全增强协议。
要点
- b2500简化协议(DWI+ADC+T1/T2)阴性预测值99.5%,适合排除病变。
- 仅DWI MIP诊断敏感度高(≥92.1%)但特异度有限(≥58.3%)。
- 简化协议与全增强协议观察者间一致性高(ICC最高0.996)。
核心结论
- 基于超高b值DWI的非增强简化乳腺MRI协议可实现与增强MRI接近的诊断准确性,适合病灶检测。
研究设计与样本
- 前瞻性研究经机构审查委员会批准,纳入127例有不同临床指征行乳腺MRI的女性。
- 使用1.5T和3T扫描仪,序列包括b值1500和2500 s/mm2的DWI、平扫T1和T2加权成像及增强序列。
- 两名阅片者独立评估,分四步读片:A为DWI最大强度投影,B为DWI+ADC图,C为B+平扫T1/T2,D为C+增强图像(全长协议)。
- 显著病变经组织学确认,不显著及明显良性病变参考完整MRI数据集。
- 统计分析包括诊断准确性计算和组内相关系数评估阅片者间一致性。
诊断标准与关键数值
- 队列包括116例BI-RADS 1类发现和138例BI-RADS≥2类发现,其中38例组织学证实为恶性。
- 步骤A(DWI MIP):阴性预测值≥97.0%,敏感度≥92.1%(两名阅片者),特异度≥58.3%,阳性预测值≥28.6%。
- 平扫读片步骤C(b2500)表现最佳:阴性预测值99.5%,敏感度97.4%,特异度88.4%。
- 组内相关系数范围0.683(MIP b1500)至0.996(全长协议)。
临床应用与阅片要点
- 超高b值DWI简化协议可用于乳腺癌筛查或病变检测,减少对比剂使用。
- b2500简化协议阴性预测值高,适合排除病变,但需注意仅DWI MIP特异度有限。
- 阅片时可结合DWI、ADC图及平扫T1/T2加权图像提高诊断准确性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
英文原文(摘要)· 点击展开
OBJECTIVES: Contrast-enhanced (CE) magnetic resonance imaging (MRI) is the most effective imaging modality for breast cancer detection. A contrast agent-free examination technique would be desirable for breast MRI screening. The purpose of this study was to evaluate the capability to detect and characterize suspicious breast lesions with an abbreviated, non-contrast-enhanced MRI protocol featuring ultra-high b-value diffusion-weighted imaging (DWI) compared with CE images.
MATERIALS AND METHODS: The institutional review board-approved prospective study included 127 female subjects with different clinical indications for breast MRI. Magnetic resonance imaging examinations included DWI sequences with b-values of 1500 s/mm2 (b1500) and 2500 s/mm2 (b2500), native T1- and T2-weighted images, and CE sequences at 1.5 T and 3 T scanners. Two reading rounds were performed, including either the b1500 or the b2500 DWI in consecutive assessment steps: (A) maximum intensity projections (MIPs) of DWI, (B) DWI and apparent diffusion coefficient maps, (C) as (B) but with additional native T1- and T2-weighted images, and (D) as (C) but with additional CE images (full-length protocol). Two readers independently determined the presence of a suspicious lesion. Histological confirmation was obtained for conspicuous lesions, whereas the full MRI data set was obtained for inconspicuous and clearly benign lesions. Statistical analysis included calculation of diagnostic accuracy and interrater agreement via the intraclass correlation coefficient.
RESULTS: The cohort comprised 116 cases with BI-RADS 1 findings and 138 cases with BI-RADS ≥2 findings, including 38 histologically confirmed malignancies. For (A), breasts without pathological findings could be recognized with high diagnostic accuracy (negative predictive value, ≥97.0%; sensitivity, ≥92.1% for both readers), but with a limited specificity (≥58.3%; positive predictive value, ≥28.6%). Within the native readings, approach (C) with b2500 performed best (negative predictive value, 99.5%; sensitivity, 97.4%; specificity, 88.4%). The intraclass correlation coefficient was between 0.683 (MIP b1500) and 0.996 (full protocol).
CONCLUSIONS: A native abbreviated breast MRI protocol with advanced high b-value DWI might allow nearly equivalent diagnostic accuracy as CE breast MRI and seems to be well suited for lesion detection purposes.
引用
- Ohlmeyer S, Laun FB, Bickelhaupt S, et al. Ultra-High b-Value Diffusion-Weighted Imaging-Based Abbreviated Protocols for Breast Cancer Detection. Invest Radiol. 2021. PMID: 34494995. doi:10.1097/RLI.0000000000000784
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。