Razaghi Ghazale, Aghsaei Fard Masoud, Hejazi Marjaneh
Medical Physics and Biomedical Engineering Department, School of Medicine, Tehran University of Medical Sciences, Tehran, Iran.
Department of Ophthalmology, Farabi Eye Hospital, Tehran University of Medical Sciences, Tehran, Iran.
J Ophthalmic Vis Res. 2023 Feb 21;18(1):41-50. doi: 10.18502/jovr.v18i1.12724. eCollection 2023 Jan-Mar.
In this study, an algorithm based on deep learning was presented to reduce the retinal nerve fiber layer (RNFL) segmentation errors in spectral domain optical coherence tomography (SD-OCT) scans using ophthalmologists' manual segmentation as a reference standard.
In this study, we developed an image segmentation network based on deep learning to automatically identify the RNFL thickness from B-scans obtained with SD-OCT. The scans were collected from Farabi Eye Hospital (500 B-scans were used for training, while 50 were used for testing). To remove the speckle noise from the images, preprocessing was applied before training, and postprocessing was performed to fill any discontinuities that might exist. Afterward, output masks were analyzed for their average thickness. Finally, the calculation of mean absolute error between predicted and ground truth RNFL thickness was performed.
Based on the testing database, SD-OCT segmentation had an average dice similarity coefficient of 0.91, and thickness estimation had a mean absolute error of 2.23 2.1 μm. As compared to conventional OCT software algorithms, deep learning predictions were better correlated with the best available estimate during the test period (r = 0.99 vs r = 0.88, respectively; 0.001).
Our experimental results demonstrate effective and precise segmentation of the RNFL layer with the coefficient of 0.91 and reliable thickness prediction with MAE 2.23 2.1 μm in SD-OCT B-scans. Performance is comparable with human annotation of the RNFL layer and other algorithms according to the correlation coefficient of 0.99 and 0.88, respectively, while artifacts and errors are evident.
在本研究中，提出了一种基于深度学习的算法，以眼科医生的手动分割作为参考标准，减少光谱域光学相干断层扫描（SD-OCT）中视网膜神经纤维层（RNFL）的分割误差。
在本研究中，我们开发了一种基于深度学习的图像分割网络，以从SD-OCT获得的B扫描中自动识别RNFL厚度。扫描数据收集自法拉比眼科医院（500次B扫描用于训练，50次用于测试）。为了去除图像中的斑点噪声，在训练前进行预处理，并进行后处理以填补可能存在的任何不连续处。之后，分析输出掩码的平均厚度。最后，计算预测的和真实的RNFL厚度之间的平均绝对误差。
基于测试数据库，SD-OCT分割的平均骰子相似系数为0.91，厚度估计的平均绝对误差为2.23±2.1μm。与传统的OCT软件算法相比，深度学习预测在测试期间与最佳可用估计的相关性更好（分别为r = 0.99和r = 0.88；P < 0.001）。
我们的实验结果表明，在SD-OCT B扫描中，RNFL层的分割有效且精确，系数为0.91，厚度预测可靠，MAE为2.23±2.1μm。根据相关系数分别为0.99和0.88，其性能与RNFL层的人工标注和其他算法相当，同时伪影和误差明显。
