Abstract: Objective To observe auto segmentation error (Seg E) determined by high definition optical coherence tomography (HD-OCT) in epiretinal membrane (ERM) with different central macular thicknesses (CMT). Design Diagnostic test. Participants Subjects were recruited from October 2018 to October 2019 in Beijing Tongren Hospital who were detected in HD-OCT and diagnosed as ERM (147 patients) and ERM with primary open-angle glaucoma (POAG) (32 patients). Methods The photographs of ocular fundus including color fundus camera, confocal scanning laser ophthalmoscope (cSLO) imaging, and HD-OCT were analyzed. The subjects were divided into A group ( ERM with CMT≤450 μm, 78 eyes), B group(ERM with CMT＞450 μm, 69 eyes), C group (ERM with POAG, CMT≤450 μm, 32 eyes). HD-OCT images were evaluated by two doctors to discriminate the accuracy of segmentation of 11 layers  macular area. The number of the auto Seg E was calculated in A, B and C group separately and compared. Main Outcome Measures The rate of Seg E of ERM. Results 94 auto Seg E of eyes were determined among 179 ERM subjects (63.95%) (A group 37.18%, B group 94.20%, C group 40.63%). The percent of Seg Es in ganglion cell layer, inner plexiform layer inner nuclear layer and outer plexiform layer in group B were significantly higher than that of group A (all P＜0.001). In addition, in other layers in group B shows no obvious differences compared with in group A (all P＞0.05). There were no significant differences in all layers between group A and group C (all P＞0.05). Conclusion The Seg E was easily found in HD-OCT scans of ERM, it was mainly located in macular inner layers. Especially when ERM with obvious macular edema, the number of Seg Es was significantly increased. (Ophthalmol CHN, 2021, 30: 343-348)

Key words: glaucoma, epiretinal membrane, automated segmentation error, optical coherence tomography