Woo Rany, Chan R V Paul, Vinekar Anand, Chiang Michael F
Department of Ophthalmology, Jules Stein Eye Institute, University of California, Los Angeles, Los Angeles, CA, USA.
Graefes Arch Clin Exp Ophthalmol. 2015 Feb;253(2):181-7. doi: 10.1007/s00417-014-2857-2. Epub 2014 Nov 21.
To evaluate aggressive posterior retinopathy of prematurity (AP-ROP) with regard to inter-expert diagnostic agreement and quantitative vascular features.
Eight ROP experts interpreted 15 retinal images for AP-ROP and plus disease. Inter-expert agreement was calculated by absolute agreement for AP-ROP and plus, and kappa statistic for each expert was compared with others. Retinal vessels were analyzed by a computer-based system to calculate diameter and integrated curvature (IC). Consensus reference standards for images were developed, and quantitative parameters for arterioles and venules were compared among images with AP-ROP vs. not AP-ROP, plus vs. not plus, and AP-ROP vs. plus.
Mean kappa for each expert in AP-ROP diagnosis ranged from -0.15 (no agreement) to 0.42 (moderate agreement). Nine (30 %) of 30 total AP-ROP diagnoses were also classified as not plus disease. Analysis of images with AP-ROP vs. plus showed that images with AP-ROP had higher venular IC (p = 0.04). Arteriolar IC was statistically significant between images with AP-ROP vs. not AP-ROP (p = 0.01) and plus vs. not plus (p = 0.00003). There were no statistically significant differences in diameter between image groups.
Inter-expert agreement with regard to AP-ROP diagnosis is imperfect. Venular curvature may be a distinguishing characteristic between AP-ROP and plus. Future studies involving quantitative features of AP-ROP will have benefits for clinical diagnosis and management.
评估早产儿侵袭性后部视网膜病变(AP-ROP)的专家间诊断一致性及定量血管特征。
八位ROP专家对15张用于诊断AP-ROP和附加病变的视网膜图像进行解读。通过计算AP-ROP和附加病变的绝对一致性来评估专家间的一致性,并将每位专家的kappa统计量与其他专家进行比较。利用基于计算机的系统分析视网膜血管,以计算血管直径和积分曲率(IC)。制定图像的共识参考标准,并比较AP-ROP与非AP-ROP、附加病变与非附加病变以及AP-ROP与附加病变图像中动静脉的定量参数。
每位专家在AP-ROP诊断中的平均kappa值范围为-0.15(无一致性)至0.42(中等一致性)。在总共30例AP-ROP诊断中,有9例(30%)也被归类为非附加病变。对AP-ROP与附加病变图像的分析表明,AP-ROP图像的静脉IC更高(p = 0.04)。在AP-ROP与非AP-ROP图像(p = 0.01)以及附加病变与非附加病变图像(p = 0.00003)之间,动脉IC有统计学显著差异。图像组之间的直径无统计学显著差异。
专家间关于AP-ROP诊断的一致性并不理想。静脉曲率可能是AP-ROP与附加病变之间的一个区别特征。未来涉及AP-ROP定量特征的研究将有助于临床诊断和管理。