Dixon 还是 DWI——比较脂肪分数与表观扩散系数在鉴别恶性与急性骨质疏松性椎体骨折中的效用
回顾性分析 42 例恶性与 27 例急性/亚急性骨质疏松性椎体骨折(38 例患者)的 1.5 T MRI,比较脂肪分数(FF)与表观扩散系数(ADC)的鉴别效能。恶性骨折的平均 FF 和 ADC 均显著低于骨质疏松性骨折,FF 的最佳阈值为 11.5%(敏感度/特异度 86%/89%),ADC 的最佳阈值为 1.04×10⁻³ mm²/s(敏感度/特异度 62%/96%)。FF 的 AUC(0.93)显著大于 ADC 的 AUC(0.82,p=0.03),作为单一鉴别指标 FF 优于 ADC。
要点
- 恶性椎体骨折的平均 FF 和 ADC 均显著低于急性/亚急性骨质疏松性骨折(p<0.001)。
- FF 最佳阈值 11.5%,鉴别恶性骨折的敏感度/特异度分别为 86%/89%,AUC 0.93。
- ADC 最佳阈值 1.04×10⁻³ mm²/s,敏感度/特异度 62%/96%,AUC 0.82,低于 FF(p=0.03)。
核心结论
- FF 和 ADC 可提高阅片者鉴别恶性与急性或亚急性骨质疏松性椎体骨折的特异度。
- 作为单一鉴别指标,FF 优于 ADC。
研究设计与样本
- 回顾性研究,纳入 42 例恶性与 27 例急性/亚急性骨质疏松性椎体骨折(38 例患者)的 1.5 T MRI。
- 两名阅片者基于常规影像形态独立将骨折分类为恶性或骨质疏松性,并评定诊断信心为有信心或无信心。
- FF 来自轴位 T1 梯度回波 2 点 Dixon MRI,ADC 图由轴位 b50 和 b900 图像计算。
- 两名阅片者独立对同一骨折椎体进行 ROI 测量平均 FF 和 ADC,比较两者数值、ROC 曲线及优化阈值效能,并以组内相关系数(ICC)分析阅片者间一致性。
关键影像表现/征象
- 恶性骨折的平均 FF 为 9.5%,骨质疏松性骨折为 32%(p<0.001)。
- 恶性骨折的平均 ADC 为 1.05×10⁻³ mm²/s,骨质疏松性骨折为 1.34×10⁻³ mm²/s(p<0.001)。
诊断标准与关键数值
- FF 最佳阈值 11.5%,鉴别恶性骨折的敏感度 86%、特异度 89%,AUC 0.93。
- ADC 最佳阈值 1.04×10⁻³ mm²/s,鉴别恶性骨折的敏感度 62%、特异度 96%,AUC 0.82。
- FF 的 AUC(0.93)显著大于 ADC 的 AUC(0.82,p=0.03)。
- 在 9 例专家阅片者信心低的亚组中,FF 优化阈值特异度 83%、ADC 优化阈值特异度 83%,均超过阅片者特异度 50%。
- 平均 FF 的阅片者间一致性 ICC=0.99(极好),平均 ADC 的 ICC=0.86(良好)。
临床应用与阅片要点
- 对常规形态学难以定性的椎体骨折,可测量平均 FF 和 ADC 辅助鉴别恶性与急性/亚急性骨质疏松性骨折。
- FF 作为单一鉴别指标优于 ADC,FF 阈值 11.5% 时敏感度 86%、特异度 89%。
- 在阅片者信心低的病例中,FF 和 ADC 优化阈值的特异度(均 83%)高于阅片者特异度(50%)。
局限性
- 原文未报告局限性。
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
PURPOSE: To compare fat fraction (FF) and apparent diffusion coefficient (ADC) as discriminators distinguishing malignant from acute/subacute osteoporotic vertebral fractures.
METHOD: 1.5 T MRIs of 42 malignant and 27 acute/subacute osteoporotic vertebral fractures (38 patients) were retrospectively reviewed. Two readers independently classified fractures as malignant or osteoporotic based on conventional imaging morphology. Diagnostic reader confidence was rated as confident or not confident. FF was derived from axial T1 gradient-echo 2-point Dixon MRI. ADC maps were calculated from axial b50 and b900 images. Both readers independently performed ROI measurements of mean FF and ADC of the same fractured vertebrae. FF and ADC values, corresponding ROC curves and optimized cut-off value performance were compared. Inter-reader agreement was analysed by calculation of intraclass correlation coefficients (ICCs). A p-value < 0.05 was deemed significant.
RESULTS: Mean FF and ADC were significantly lower in malignant (9.5 % and 1.05 × 10-3 mm²/s) compared to osteoporotic fractures (32 % and 1.34 × 10-3 mm²/s, all p < 0.001). The optimal cut-off FF was 11.5 %, detecting malignant fractures with 86 %/89 % sensitivity/specificity. The optimal ADC cut-off of 1.04 × 10-3 mm/s² yielded 62 %/96 % sensitivity/specificity. FF AUC (0.93) was significantly larger than ADC AUC (0.82, p = 0.03). In the subgroup of nine cases reported with low expert reader confidence, the optimized cut-off specificities of FF (83 %) and ADC (83 %) exceeded reader specificity (50 %). There was excellent inter-reader agreement for mean FF (ICC = 0.99) and good agreement for mean ADC (ICC = 0.86) measurements.
CONCLUSION: FF and ADC can improve reader specificity to distinguish between malignant and acute or subacute osteoporotic vertebral fractures. As single discriminator, FF was superior to ADC.
引用
- Donners R, Obmann MM, Boll D, et al. Dixon or DWI - Comparing the utility of fat fraction and apparent diffusion coefficient to distinguish between malignant and acute osteoporotic vertebral fractures. Eur J Radiol. 2020. PMID: 33068837. doi:10.1016/j.ejrad.2020.109342
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。