Midroni Julie, Longwell Jack, Bhambra Nishaant, Demian Sueellen, Pecaku Aurora, Martins Melo Isabela, Muni Rajeev H
Department of Medicine, Temerty Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada.
Temerty Centre for Artificial Intelligence Research and Education in Medicine, Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, Ontario, Canada.
Ophthalmol Sci. 2025 Jun 16;5(6):100852. doi: 10.1016/j.xops.2025.100852. eCollection 2025 Nov-Dec.
We present an algorithm to segment subretinal fluid (SRF) on individual B-scan slices in patients with rhegmatogenous retinal detachment (RRD). Particular attention is paid to robustness, with a fivefold cross-validation approach and a hold-out test set.
Retrospective, cross-sectional study.
A total of 3819 B-scan slices across 98 time points from 45 patients were used in this study.
Subretinal fluid was segmented on all scans. A base SegFormer model, pretrained on 4 massive data sets, was further trained on raw B-scans from the retinal OCT fluid challenge data set of 4532 slices: an open data set of intraretinal fluid, SRF, and pigment epithelium detachment. When adequate performance was reached, transfer learning was used to train the model on our in-house data set, to segment SRF by generating a pixel-wise mask of presence/absence of SRF. A fivefold cross-validation approach was used, with an additional hold-out test set. All folds were first trained and cross-validated and then additionally tested on the hold-out set. Mean (averaged across images) and total (summed across all pixels, irrespective of image) Dice coefficients were calculated for each fold.
Subretinal fluid volume after surgical intervention for RRD.
The average total Dice coefficient across the validation folds was 0.92, the average mean Dice coefficient was 0.82, and the median Dice was 0.92. For the test set, the average total Dice coefficient was 0.94, the average mean Dice coefficient was 0.82, and the median Dice was 0.92. The model showed strong interfold consistency on the hold-out set, with a standard deviation of only 0.03.
The SegFormer model for SRF segmentation demonstrates a strong ability to segment SRF. This result holds up to cross-validation and hold-out testing, across all folds. The model is available open-source online.
Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.
我们提出一种算法，用于对孔源性视网膜脱离（RRD）患者的个体B扫描切片上的视网膜下液（SRF）进行分割。特别关注稳健性，采用五重交叉验证方法和一个留出测试集。
回顾性横断面研究。
本研究使用了来自45名患者的98个时间点的总共3819个B扫描切片。
对所有扫描进行视网膜下液分割。一个在4个大规模数据集上预训练的基础SegFormer模型，在来自4532个切片的视网膜OCT液体挑战数据集的原始B扫描上进一步训练：一个关于视网膜内液、SRF和色素上皮脱离的开放数据集。当达到足够的性能时，使用迁移学习在我们的内部数据集上训练模型，通过生成SRF存在/不存在的逐像素掩码来分割SRF。采用五重交叉验证方法，还有一个额外的留出测试集。所有折叠首先进行训练和交叉验证，然后在留出集上进行额外测试。为每个折叠计算平均（跨图像平均）和总计（跨所有像素求和，不考虑图像）的Dice系数。
RRD手术干预后的视网膜下液体积。
验证折叠的平均总计Dice系数为0.92，平均平均Dice系数为0.82，中位数Dice为0.92。对于测试集，平均总计Dice系数为0.94，平均平均Dice系数为0.82，中位数Dice为0.92。该模型在留出集上显示出很强的折叠间一致性，标准差仅为0.03。
用于SRF分割的SegFormer模型展示了很强的SRF分割能力。这一结果在所有折叠的交叉验证和留出测试中都成立。该模型可在网上开源获取。
在本文末尾的脚注和披露中可能会找到专有或商业披露信息。
