Tomaru Y, Inoue T, Oriuchi N, Takahashi K, Endo K
Department of Nuclear Medicine, Gunma University School of Medicine, Gunma, Japan.
Eur J Nucl Med. 1998 Jan;25(1):55-9.
In calculating the relative and absolute renal uptake of technetium-99m mercaptoacetyltriglycine (MAG3), inter-operator variability in the assignment of the renal region of interest (ROI) is a critical factor. Our goal was to develop a semi-automated method of assigning the renal ROI and then to compare the inter-operator variability in calculating the percent injected dose (%ID) in the kidney at 1-2 min, using semi-automated versus manual ROIs. The manual ROIs were drawn independently by three operators (A, B and C). Operator A had about 20 years, experience in nuclear medicine, while operators B and C respectively had 3 years and 1 year of experience. In the semi-automated renal ROI selection method using the double-threshold technique, the operators only click around the centre of each kidney. The same three operators processed the ROIs using this double-threshold method on 1-2 min images. The semi-automated method failed in three kidneys with very markedly reduced function owing to superimposition by liver or spleen. Inter-operator reproducibility in the remaining 59 kidneys was estimated using manual and semi-automated ROIs. With manual ROIs, the %ID (mean+/-standard error of mean) was 4.32+/-0.167 for A, 4. 14+/-0.165 for B and 3.28+/-0.139 for C. Although there was good correlation among them, these values were significantly different (P<0.0001). Using semi-automated ROIs, the %ID was 4.38+/-0.160 for three operators. No significant difference was observed. Complete reproducibility was shown in 58 of 59 kidneys; the %ID difference of the remaining kidney was only 1.2%. The lowest %ID of all the kidneys successfully detected using the semi-automated method was 0. 77%. The semi-automated renal ROI selection method using the double-threshold technique displays good detectability of the renal contour. The renal uptake calculated using this method is reproducible and acceptable in routine clinical practice.
在计算99m锝-巯基乙酰三甘氨酸（MAG3）的相对和绝对肾脏摄取量时，操作者在确定肾脏感兴趣区（ROI）方面的变异性是一个关键因素。我们的目标是开发一种半自动确定肾脏ROI的方法，然后比较使用半自动ROI与手动ROI在1 - 2分钟时计算肾脏注射剂量百分比（%ID）的操作者间变异性。手动ROI由三名操作者（A、B和C）独立绘制。操作者A有大约20年核医学经验，而操作者B和C分别有3年和1年经验。在使用双阈值技术的半自动肾脏ROI选择方法中，操作者只需在每个肾脏中心周围点击。同样的三名操作者在1 - 2分钟图像上使用这种双阈值方法处理ROI。由于肝脏或脾脏的重叠，半自动方法在三个功能明显降低的肾脏中失败。使用手动和半自动ROI估计其余59个肾脏的操作者间可重复性。对于手动ROI，A的%ID（平均值±平均标准误差）为4.32±0.167，B为4.14±0.165，C为3.28±0.139。尽管它们之间有良好的相关性，但这些值有显著差异（P<0.0001）。使用半自动ROI时，三名操作者的%ID为4.38±0.160。未观察到显著差异。59个肾脏中有58个显示出完全可重复性；其余一个肾脏的%ID差异仅为1.2%。使用半自动方法成功检测到的所有肾脏中最低的%ID为0.77%。使用双阈值技术的半自动肾脏ROI选择方法显示出良好的肾脏轮廓可检测性。使用该方法计算的肾脏摄取量在常规临床实践中是可重复且可接受的。
Zotero 插件