Legacy Devers Eye Institute, Portland, OR.
J Glaucoma. 2019 May;28(5):368-374. doi: 10.1097/IJG.0000000000001222.
PRéCIS:: There are errors in automated segmentation of the retinal nerve fiber layer (RNFL) in glaucoma suspects or patients with mild glaucoma that appear to persist over time; however, automated segmentation has greater repeatability than manual segmentation.
To identify whether optical coherence tomography (OCT) segmentation errors in RNFL thickness measurements persist longitudinally.
This was a cohort study. We used spectral domain OCT (Spectralis) to measure RNFL thickness in a 6-degree peripapillary circle, and exported the native "automated segmentation only" results. In addition, we exported RNFL thickness results after "manual refinement" to correct errors in the automated segmentation, and used the differences in these measurements as "error" in segmentation. We used Bland-Altman plots and linear regression to determine the magnitude, location, and repeatability of RNFL thickness error in all twelve 30-degree sectors and compared the error at baseline to follow-up time points at 6 months, 2 years, 3 years, and 4 years.
We included 406 eyes from 213 participants. The 95% confidence interval for errors at baseline was -6.5 to +13.2 μm. The correlation between the baseline error and the errors in the follow-up time periods were high (r>0.5, P<0.001 for all). Automated segmentation had a smaller SD of residuals from the longitudinal trend line when compared to manual refinement (1.56 vs. 1.80 μm, P<0.001), and a higher ability (P=0.009) to monitor progression using an analysis of a longitudinal signal-to-noise ratio.
Errors in automated segmentation remain relatively stable, and baseline error is highly likely to persist in the same direction and magnitude in subsequent time periods. However, automated segmentation (without manual refinement) is more repeatable and may be more sensitive to glaucomatous progression. Future segmentation algorithms could exploit these findings to improve automated segmentation in the future.
在青光眼疑似患者或轻度青光眼患者中，自动视网膜神经纤维层（RNFL）分割存在错误，这些错误似乎随着时间的推移而持续存在；然而，自动分割比手动分割具有更高的可重复性。
确定 RNFL 厚度测量的光学相干断层扫描（OCT）分割误差是否会随时间纵向持续存在。
这是一项队列研究。我们使用谱域 OCT（Spectralis）测量 6 度视盘周围的 RNFL 厚度，并导出原始的“仅自动分割”结果。此外，我们还导出了经过“手动细化”以纠正自动分割错误的 RNFL 厚度结果，并使用这些测量值的差异作为分割中的“误差”。我们使用 Bland-Altman 图和线性回归来确定所有 12 个 30 度扇形区中 RNFL 厚度误差的大小、位置和可重复性，并将基线时的误差与 6 个月、2 年、3 年和 4 年的随访时间点进行比较。
我们纳入了 213 名参与者的 406 只眼。基线时误差的 95%置信区间为-6.5 至+13.2μm。基线误差与随访时间段内的误差之间的相关性很高（所有 r>0.5，P<0.001）。与手动细化相比，自动分割具有更小的从纵向趋势线的残差标准差（1.56 与 1.80μm，P<0.001），并且使用纵向信噪比分析监测进展的能力更高（P=0.009）。
自动分割中的误差相对稳定，并且后续时间段内基线误差很可能以相同的方向和幅度持续存在。然而，自动分割（未经手动细化）具有更高的可重复性，并且可能对青光眼进展更敏感。未来的分割算法可以利用这些发现来提高未来的自动分割性能。
