Eye Research Centre, Five Senses Health Institute, and Iran University of Medical Sciences, Tehran, Iran.
Eye Research Centre, Five Senses Health Institute, and Iran University of Medical Sciences, Tehran, Iran; Stem Cell and Regenerative Medicine Research Centre, Iran University of Medical Sciences, Tehran, Iran.
Can J Ophthalmol. 2023 Dec;58(6):577-581. doi: 10.1016/j.jcjo.2022.06.013. Epub 2022 Jul 19.
To evaluate the accuracy of peripapillary optical coherence tomography angiography (OCTA) segmentation in eyes with acute nonarteritic anterior ischemic optic neuropathy (NAION) and healthy eyes.
In this retrospective study, en face OCTA images of the optic disc of healthy eyes and eyes with unilateral acute NAION were obtained. The disc boundary and radial peripapillary capillary (RPC) segmentation were generated automatically by the instrument software and then corrected by 2 expert investigators. The frequency of segmentation errors and its impact on vessel density and nerve fibre layer (NFL) thickness measurements were evaluated.
Thirty-eight eyes of 38 subjects (18 in the healthy group and 20 in the acute NAION group) were studied. A misidentified disc border was noted in 5 healthy eyes (27.7%) and 19 eyes with NAION (95.0%; p < 0.001). Segmentation error at the RPC level was found in 6 healthy eyes (33.33%) and 19 eyes with NAION (95.0%; p < 0.001). The nerve fibre layer thickness and RPC density did not change statistically significantly after error corrections in both groups.
Misidentification of disc border and segmentation error of the RPC layer are common in OCTA images of the optic disc. Accuracy of OCTA imaging in disc boundary detection and RPC network segmentation is reduced in edematous optic discs following acute NAION.
评估急性非动脉炎性前部缺血性视神经病变(NAION)患眼和健康眼的周边视网膜光相干断层扫描血管造影(OCTA)分割的准确性。
在这项回顾性研究中,获得了健康眼和单侧急性 NAION 患眼的视盘表面 OCTA 图像。仪器软件自动生成视盘边界和放射状周边毛细血管(RPC)分割,然后由 2 名专家调查员进行校正。评估分割错误的频率及其对血管密度和神经纤维层(NFL)厚度测量的影响。
研究了 38 只眼的 38 名受试者(健康组 18 只眼,急性 NAION 组 20 只眼)。在 5 只健康眼(27.7%)和 19 只 NAION 眼中发现了错误识别的视盘边界(p<0.001)。在 6 只健康眼(33.33%)和 19 只 NAION 眼中发现了 RPC 水平的分割错误(p<0.001)。两组校正后,神经纤维层厚度和 RPC 密度均无统计学显著变化。
在视盘 OCTA 图像中,视盘边界的错误识别和 RPC 层的分割错误很常见。在急性 NAION 后水肿视盘,OCTA 成像在检测视盘边界和 RPC 网络分割的准确性降低。