Ghazi Nicola G, Kirk Tyler, Allam Souha, Yan Guofen
Department of Ophthalmology, The University of Virginia Health System, Charlottesville, Virginia 22908, USA.
Am J Ophthalmol. 2009 Jul;148(1):90-6.e2. doi: 10.1016/j.ajo.2009.02.017. Epub 2009 Apr 29.
To assess error indicators encountered during optical coherence tomography (OCT) automated retinal thickness measurement (RTM) in neovascular age-related macular degeneration (NVAMD) before and after bevacizumab (Avastin; Genentech Inc, South San Francisco, California, USA) treatment.
Retrospective observational cross-sectional study.
Each of the 6 radial lines of a single Stratus fast macular OCT study before and 3 months following initiation of treatment in 46 eyes with NVAMD, for a total of 552 scans, was evaluated. Error frequency was analyzed relative to the presence of intraretinal, subretinal (SR), and subretinal pigment epithelial (SRPE) fluid. In scans with edge detection kernel (EDK) misplacement, manual caliper measurement of the central macular (CMT) and central foveal (CFT) thicknesses was performed and compared to the software-generated values. The frequency of the various types of error indicators, the risk factors for error, and the magnitude of automated RTM error were analyzed.
Error indicators were found in 91.3% and 71.7% of eyes before and after treatment, respectively (P = .013). Suboptimal signal strength was the most common error indicator. EDK misplacement was the second most common type of error prior to treatment and the least common after treatment (P = .005). Eyes with SR or SRPE fluid were at the highest risk for error, particularly EDK misplacement (P = .039). There was a strong association between the software-generated and caliper-generated CMT and CFT measurements. The software overestimated measurements by up to 32% and underestimated them by up to 15% in the presence of SR and SRPE fluid, respectively.
OCT errors are very frequent in NVAMD. SRF is associated with the highest risk and magnitude of error in automated CMT and CFT measurements. Manually adjusted measurements may be more reliable in such eyes.
评估在贝伐单抗（阿瓦斯汀；基因泰克公司，美国加利福尼亚州南旧金山）治疗前后，新生血管性年龄相关性黄斑变性（NVAMD）患者进行光学相干断层扫描（OCT）自动视网膜厚度测量（RTM）时遇到的误差指标。
回顾性观察性横断面研究。
对46只NVAMD患眼在治疗开始前及治疗3个月后的单次Stratus快速黄斑OCT研究中的6条径向线进行评估，共552次扫描。分析误差频率与视网膜内、视网膜下（SR）和视网膜色素上皮下（SRPE）液的存在情况的关系。在边缘检测内核（EDK）位置错误的扫描中，对中心黄斑（CMT）和中心凹（CFT）厚度进行手动卡尺测量，并与软件生成的值进行比较。分析各种类型误差指标的频率、误差的危险因素以及自动RTM误差的大小。
治疗前和治疗后分别有91.3%和71.7%的患眼发现误差指标（P = 0.013）。信号强度欠佳是最常见的误差指标。EDK位置错误是治疗前第二常见的误差类型，治疗后最不常见（P = 0.005）。有SR或SRPE液的患眼误差风险最高，尤其是EDK位置错误（P = 0.039）。软件生成的CMT和CFT测量值与卡尺测量值之间存在很强的相关性。在存在SR和SRPE液的情况下，软件测量值分别高估多达32%和低估多达15%。
NVAMD中OCT误差非常常见。SRF与自动CMT和CFT测量中最高的误差风险和误差大小相关。在此类患眼中，手动调整测量可能更可靠。
