Wulff A M, Bolte H, Fischer S, Freitag-Wolf S, Soza G, Tietjen C, Biederer J, Heller M, Fabel M
Klinik für Diagnostische Radiologie, Universitätsklinikum Schleswig-Holstein, Campus Kiel.
Rofo. 2012 Sep;184(9):820-8. doi: 10.1055/s-0032-1312860. Epub 2012 Aug 7.
To investigate measurement accuracy in terms of precision and inter-rater variability in the simultaneous volumetric assessment of lung, liver and lymph node metastasis size change over time in comparison to RECIST 1.1.
Three independent readers evaluated multislice CT data from clinical follow-up studies (chest/abdomen) in 50 patients with metastases. A total of 117 lung, 77 liver and 97 lymph node metastases were assessed manually (RECIST 1.1) and by volumetry with semi-automated software. The quality of segmentation and need for manual adjustments were recorded. Volumes were converted to effective diameters to allow comparison to RECIST. For statistical assessment of precision and interobserver agreement, the Wilcoxon-signed rank test and Bland-Altman plots were utilized.
The quality of segmentation after manual correction was acceptable to excellent in 95 % of lesions and manual corrections were applied in 21 - 36 % of all lesions, most predominantly in lymph nodes. Mean precision was 2.6 - 6.3 % (manual) with 0.2 - 1.5 % (effective) relative measurement deviation (p <.001). Inter-reader median variation coefficients ranged from 9.4 - 12.8 % (manual) and 2.9 - 8.2 % (volumetric) for different lesion types (p < .001). The limits of agreement were ± 9.8 to ± 11.2 % for volumetric assessment.
Superior precision and inter-rater variability of volumetric over manual measurement of lesion change over time was demonstrated in a whole body setting.
与RECIST 1.1相比，研究在同时进行肺、肝和淋巴结转移灶大小随时间变化的体积评估时，测量精度方面的精密度和观察者间变异性。
三名独立阅片者评估了50例有转移灶患者临床随访研究（胸部/腹部）的多层CT数据。总共对117个肺转移灶、77个肝转移灶和97个淋巴结转移灶进行了手动评估（RECIST 1.1），并使用半自动软件进行体积测量。记录分割质量和手动调整的必要性。将体积转换为有效直径以便与RECIST进行比较。为了对精密度和观察者间一致性进行统计学评估，采用了Wilcoxon符号秩检验和Bland-Altman图。
在95%的病灶中，手动校正后的分割质量为可接受至优秀，在所有病灶的21%-36%中进行了手动校正，最主要是在淋巴结中。平均精密度为2.6%-6.3%（手动），相对测量偏差为0.2%-1.5%（有效）（p<.001）。不同病灶类型的阅片者间中位数变异系数范围为9.4%-12.8%（手动）和2.9%-8.2%（体积测量）（p<.001）。体积评估的一致性界限为±9.8%至±11.2%。
在全身检查中，与手动测量相比，体积测量在病灶随时间变化的测量方面具有更高的精密度和观察者间变异性。
