Beslic Nermina, Milardovic Renata, Sadija Amera, Dzananovic Lejla, Cavaljuga Semra
Clinic of Nuclear Medicine, University-Clinical Center Sarajevo, Sarajevo, BiH.
Department of Epidemiology and Biostatistics, Faculty of Medicine, University of Sarajevo, Sarajevo, BiH.
Acta Inform Med. 2017 Mar;25(1):28-33. doi: 10.5455/aim.2017.25.28-33.
This study objective was to evaluate interobserver agreement between individual pairs of three nuclear medicine physicians in interpretation of renal cortical scintigraphy in children with respect to the mode of acquisition (planar vs. SPECT), diagnoses and kidney site (left vs. right).
Thirty children were imaged in planar and SPECT mode per protocol upon the injection of Tc-99m DMSA dose adjusted to their body weight. Patients were classified according to diagnoses into four groups. Three nuclear medicine physicians interpreted the findings blindly and independently. Renal defects were interpreted as focal and diffuse, per three renal segments. For the raters we calculated simple percentage agreement, the Cohen kappa statistic with 95% confidence intervals, and the overall kappa defining the levels of reliability as almost perfect or perfect, substantial, moderate, fair and slight agreement.
Interobserver agreement in planar interpretation was 77,2% (kappa=0.59; 95% confidence interval, 0.41 to 0.75) and SPECT 72,9% (kappa= 0,57; 95% confidence interval, 0,41 to 0,72). In planar interpretation, all individual pairs had moderate agreements except one that had a substantial agreement. In SPECT, all the pairs had moderate agreements except one that had an almost perfect agreement. Overall agreement per kidney site was on planar 73,4% for the left (kappa=0,54, moderate agreement), and 81,1% for the right kidney (kappa 0,63, substantial agreement). On SPECT, there was 72,2% agreement for the left (kappa=0,59, mode rate agreement), and 73,7% for the right kidney (kappa=0,54, moderate agreement). Overall agreement per diagnoses ranged from 70-88,9% on planar (kappa= -0,04 to 0,79), and 50-100% on SPECT (kappa=-0,02-1,000) indicating agreements from slight to substantial.
Our results suggest acceptable levels of interobserver agreement in all individual pairs of raters with respect to the mode of acquisition (planar vs. SPECT), diagnoses and kidney site (left vs. right). For the mode of acquisition, we would recommend hybrid imaging SPECT/CT method to be used whenever possible in the detection of renal cortical defects on Tc-99m-DMSA scintigraphy.
本研究旨在评估三位核医学医师在解读儿童肾皮质闪烁显像时,在采集方式(平面显像与单光子发射计算机断层扫描[SPECT])、诊断及肾脏部位(左肾与右肾)方面的观察者间一致性。
按照方案,对30名儿童在注射根据体重调整剂量的锝-99m二巯丁二酸(Tc-99m DMSA)后进行平面显像和SPECT显像。根据诊断将患者分为四组。三位核医学医师独立且盲法解读检查结果。根据三个肾段将肾缺损解读为局灶性和弥漫性。对于评估者,我们计算了简单百分比一致性、95%置信区间的Cohen κ统计量,以及将可靠性水平定义为几乎完美或完美、实质性、中等、一般和轻微一致性的总体κ统计量。
平面显像的观察者间一致性为77.2%(κ=0.59;95%置信区间为0.41至0.75),SPECT显像为72.9%(κ=0.57;95%置信区间为0.41至0.72)。在平面显像解读中,除一组为实质性一致性外,所有评估者对之间均为中等一致性。在SPECT显像中,除一组为几乎完美一致性外,所有对之间均为中等一致性。每个肾脏部位的总体一致性在平面显像中,左肾为73.4%(κ=0.54,中等一致性),右肾为81.1%(κ=0.63,实质性一致性)。在SPECT显像中,左肾一致性为72.2%(κ=0.59,中度一致性),右肾为73.7%(κ=0.54,中等一致性)。每个诊断的总体一致性在平面显像中为70 - 88.9%(κ=-0.04至0.79),在SPECT显像中为50 - 100%(κ=-0.02至1.000),表明一致性从轻微到实质性。
我们的结果表明,所有评估者对在采集方式(平面显像与SPECT)、诊断及肾脏部位(左肾与右肾)方面的观察者间一致性处于可接受水平。对于采集方式,我们建议在通过Tc-99m-DMSA闪烁显像检测肾皮质缺损时,尽可能使用SPECT/CT混合成像方法。