Department of Statistics and Data Science, National University of Singapore, Singapore.
Ophthalmic Engineering & Innovation Laboratory, Singapore Eye Research Institute, Singapore National Eye Centre, Singapore.
Transl Vis Sci Technol. 2023 Feb 1;12(2):23. doi: 10.1167/tvst.12.2.23.
(1) To assess the performance of geometric deep learning in diagnosing glaucoma from a single optical coherence tomography (OCT) scan of the optic nerve head and (2) to compare its performance to that obtained with a three-dimensional (3D) convolutional neural network (CNN), and with a gold-standard parameter, namely, the retinal nerve fiber layer (RNFL) thickness.
Scans of the optic nerve head were acquired with OCT for 477 glaucoma and 2296 nonglaucoma subjects. All volumes were automatically segmented using deep learning to identify seven major neural and connective tissues. Each optic nerve head was then represented as a 3D point cloud with approximately 1000 points. Geometric deep learning (PointNet) was then used to provide a glaucoma diagnosis from a single 3D point cloud. The performance of our approach (reported using the area under the curve [AUC]) was compared with that obtained with a 3D CNN, and with the RNFL thickness.
PointNet was able to provide a robust glaucoma diagnosis solely from a 3D point cloud (AUC = 0.95 ± 0.01).The performance of PointNet was superior to that obtained with a 3D CNN (AUC = 0.87 ± 0.02 [raw OCT images] and 0.91 ± 0.02 [segmented OCT images]) and with that obtained from RNFL thickness alone (AUC = 0.80 ± 0.03).
We provide a proof of principle for the application of geometric deep learning in glaucoma. Our technique requires significantly less information as input to perform better than a 3D CNN, and with an AUC superior to that obtained from RNFL thickness.
Geometric deep learning may help us to improve and simplify diagnosis and prognosis applications in glaucoma.
（1）评估从视神经头的单次光学相干断层扫描（OCT）中进行几何深度学习诊断青光眼的性能，以及（2）比较其与三维（3D）卷积神经网络（CNN）的性能，以及与金标准参数，即视网膜神经纤维层（RNFL）厚度的比较。
使用 OCT 采集了 477 例青光眼和 2296 例非青光眼患者的视神经头扫描。所有体积均使用深度学习自动分割，以识别七种主要的神经和连接组织。然后，每个视神经头都表示为具有大约 1000 个点的 3D 点云。然后使用几何深度学习（PointNet）从单个 3D 点云中提供青光眼诊断。我们的方法（通过曲线下面积[AUC]报告）的性能与 3D CNN 以及 RNFL 厚度的性能进行了比较。
PointNet 能够仅从 3D 点云中提供稳健的青光眼诊断（AUC = 0.95 ± 0.01）。PointNet 的性能优于 3D CNN（AUC = 0.87 ± 0.02[原始 OCT 图像]和 0.91 ± 0.02[分割 OCT 图像]）和单独来自 RNFL 厚度的性能（AUC = 0.80 ± 0.03）。
我们提供了几何深度学习在青光眼应用中的原理证明。我们的技术需要输入的信息量明显减少，性能优于 3D CNN，并且 AUC 优于来自 RNFL 厚度的 AUC。
我们提供了几何深度学习在青光眼应用中的原理证明。与 3D CNN 相比，我们的技术需要输入的信息量明显减少，性能更好，并且 AUC 优于来自 RNFL 厚度的 AUC。
原文中出现的“OCT”和“CNN”分别指“光学相干断层扫描”和“卷积神经网络”，在翻译时保留了英文缩写。
