Provenzale James M, Taylor Brian A, Wilde Elisabeth A, Boss Michael, Schneider Walter
1 Box 3808, Department of Radiology, Duke University Medical Center, Durham, NC 27516, USA.
2 Department of Biomedical Engineering, College of Engineering, Virginia Commonwealth University, USA.
Neuroradiol J. 2018 Dec;31(6):581-586. doi: 10.1177/1971400918789383. Epub 2018 Jul 24.
We employed a novel diffusion tensor imaging phantom to study intra- and interscanner reproducibility on two 3T magnetic resonance (MR) scanners. Using a phantom containing thousands of hollow micron-size tubes in complex arrays, we performed two experiments using a b value of 1000 s/ms on two Siemens 3T Trio scanners. First, we performed 12-direction scans. Second, on one scanner, we performed two 64-direction protocols with different repetition times (TRs). We used a one-way analysis of variance to calculate differences between scanners and the Mann-Whitney U test to assess differences between 12-direction and 64-direction data. We calculated the coefficient of variation (CoV) for intrascanner and interscanner data. For 12-direction protocols, mean fractional anisotropy (FA) was 0.3003 for Scanner 1 (four scans) and 0.3094 for Scanner 2 (three scans). Lowest FA value on Scanner 1 was 2.56 standard deviations below the mean of Scanner 2. For 64-direction scans, mean FA was 0.2640 for 4000 ms TR and 0.2582 for 13,200 ms TR scans. For 12-direction scans, within-scanner CoV was 0.0326 for Scanner 1 and 0.0240 for Scanner 2; between-scanner CoV was 0.032. For 64-direction scans, CoV was 0.056 for TR 4000 ms and 0.0533 for TR 13,200 ms. The difference between median FA values of 12-direction and 64-direction scans was statistically significant ( p < 0.001). We found relatively good reproducibility on any single MR scanner. FA values from one scanner were sometimes significantly below the mean FA of another scanner, which has important implications for clinical use of DTI.
我们使用了一种新型扩散张量成像体模，来研究两台3T磁共振（MR）扫描仪上的扫描内和扫描仪间的可重复性。我们使用了一个包含数千个复杂排列的中空微米级管道的体模，在两台西门子3T Trio扫描仪上，使用b值为1000 s/ms进行了两项实验。首先，我们进行了12方向扫描。其次，在一台扫描仪上，我们使用不同的重复时间（TR）进行了两个64方向的扫描方案。我们使用单向方差分析来计算扫描仪之间的差异，并使用曼-惠特尼U检验来评估12方向和64方向数据之间的差异。我们计算了扫描内和扫描仪间数据的变异系数（CoV）。对于12方向扫描方案，扫描仪1（四次扫描）的平均分数各向异性（FA）为0.3003，扫描仪2（三次扫描）为0.3094。扫描仪1上的最低FA值比扫描仪2的平均值低2.56个标准差。对于64方向扫描，4000 ms TR的平均FA为0.2640，13200 ms TR扫描为0.2582。对于12方向扫描，扫描仪1的扫描内CoV为0.0326，扫描仪2为0.0240；扫描仪间CoV为0.032。对于64方向扫描，4000 ms TR的CoV为0.056，13200 ms TR的CoV为0.0533。12方向和64方向扫描的FA中位数差异具有统计学意义（p < 0.001）。我们发现在任何一台单一的MR扫描仪上都具有相对良好的可重复性。来自一台扫描仪的FA值有时显著低于另一台扫描仪的平均FA值，这对DTI的临床应用具有重要意义。
