数字乳腺X线摄影与数字乳腺断层摄影中的自动乳腺密度计算:对平均腺体剂量和BIRADS密度分类的影响
研究选取561例无乳腺病变患者的全视野数字乳腺X线摄影(FFDM)和数字乳腺断层摄影(DBT)检查,用商业软件Quantra 2D和Quantra 3D分别自动计算容积乳腺密度。两种算法测量结果相关性良好(r≥0.77),但多数参数中位数存在统计学显著差异(P<0.05),FFDM的容积乳腺密度和面积乳腺密度中位数分别比DBT估计值高8%和77%。两种算法对BIRADS分类和平均腺体剂量的临床结局无差异。
要点
- FFDM的容积乳腺密度中位数比DBT高8%,面积乳腺密度中位数比DBT高77%。
- 两种算法将35%和55%的乳腺分别归入BIRADS b和c类。
- 致密乳腺(BIRADS d类)中FFDM与DBT的平均腺体剂量中位数相似。
核心结论
- 两种算法的乳腺密度测量导致等效的BIRADS分类和平均腺体剂量值,临床结局无差异;致密乳腺(BIRADS d类)中FFDM与DBT的平均腺体剂量中位数相似。
研究设计与样本
- 研究纳入561例无乳腺病变患者的FFDM和DBT检查,使用商业软件Quantra 2D和Quantra 3D分别自动计算容积乳腺密度,并评估面积乳腺密度和总乳腺体积;采用Mann-Whitney U非参数检验和Spearman秩相关系数进行数据相关性分析,平均腺体剂量按Dance等提出的方法计算。
诊断标准与关键数值
- 两种算法测量结果相关性良好(r≥0.77)。
- FFDM的容积乳腺密度中位数比DBT估计值高8%,面积乳腺密度中位数比DBT估计值高77%。
- 两种算法将35%的乳腺归入BIRADS b类,55%归入BIRADS c类。
- DBT的平均腺体剂量高于FFDM,BIRADS d类乳腺中差异较小(5%)。
- 平均腺体剂量平均比Dance等提出的乳腺腺体含量方法所得值高6%。
临床应用与阅片要点
- 原文未据此给出推荐。
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
RATIONALE AND OBJECTIVES: The study aimed to compare the breast density estimates from two algorithms on full-field digital mammography (FFDM) and digital breast tomosynthesis (DBT) and to analyze the clinical implications.
MATERIALS AND METHODS: We selected 561 FFDM and DBT examinations from patients without breast pathologies. Two versions of a commercial software (Quantra 2D and Quantra 3D) calculated the volumetric breast density automatically in FFDM and DBT, respectively. Other parameters such as area breast density and total breast volume were evaluated. We compared the results from both algorithms using the Mann-Whitney U non-parametric test and the Spearman's rank coefficient for data correlation analysis. Mean glandular dose (MGD) was calculated following the methodology proposed by Dance et al.
RESULTS: Measurements with both algorithms are well correlated (r ≥ 0.77). However, there are statistically significant differences between the medians (P < 0.05) of most parameters. The volumetric and area breast density median values from FFDM are, respectively, 8% and 77% higher than DBT estimations. Both algorithms classify 35% and 55% of breasts into BIRADS (Breast Imaging-Reporting and Data System) b and c categories, respectively. There are no significant differences between the MGD calculated using the breast density from each algorithm. DBT delivers higher MGD than FFDM, with a lower difference (5%) for breasts in the BIRADS d category. MGD is, on average, 6% higher than values obtained with the breast glandularity proposed by Dance et al.
CONCLUSIONS: Breast density measurements from both algorithms lead to equivalent BIRADS classification and MGD values, hence showing no difference in clinical outcomes. The median MGD values of FFDM and DBT examinations are similar for dense breasts (BIRADS d category).
引用
- Castillo-García M, Chevalier M, Garayoa J, et al. Automated Breast Density Computation in Digital Mammography and Digital Breast Tomosynthesis: Influence on Mean Glandular Dose and BIRADS Density Categorization. Acad Radiol. 2017. PMID: 28214227. doi:10.1016/j.acra.2017.01.011
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。