Roski Eye Institute, Department of Ophthalmology, Keck School of Medicine at the University of Southern California, Los Angeles, CA, USA.
Sol Price School of Public Policy, University of Southern California, Los Angeles, CA, USA.
Transl Vis Sci Technol. 2020 Mar 30;9(2):18. doi: 10.1167/tvst.9.2.18. eCollection 2020 Mar.
To develop a deep neural network that detects the scleral spur in anterior segment optical coherence tomography (AS-OCT) images.
Participants in the Chinese American Eye Study, a population-based study in Los Angeles, California, underwent complete ocular examinations, including AS-OCT imaging with the Tomey CASIA SS-1000. One human expert grader provided reference labels of scleral spur locations in all images. A convolutional neural network (CNN)-based on the ResNet-18 architecture was developed to detect the scleral spur in each image. Performance of the CNN model was assessed by calculating prediction errors, defined as the difference between the Cartesian coordinates of reference and CNN-predicted scleral spur locations. Prediction errors were compared with intragrader variability in detecting scleral spur locations by the reference grader.
The CNN was developed using a training dataset of 17,704 images and tested using an independent dataset of 921 images. The mean absolute prediction errors of the CNN model were 49.27 ± 42.07 µm for X-coordinates and 47.73 ± 39.70 µm for Y-coordinates. The mean absolute intragrader variability was 52.31 ± 47.75 µm for X-coordinates and 45.88 ± 45.06 µm for Y-coordinates. Distributions of prediction errors for the CNN and intragrader variability for the reference grader were similar for X-coordinates ( = 0.609) and Y-coordinates ( = 0.378). The mean absolute prediction error of the CNN was 73.08 ± 52.06 µm and the mean absolute intragrader variability was 73.92 ± 60.72 µm.
A deep neural network can detect the scleral spur on AS-OCT images with performance similar to that of a human expert grader.
Deep learning methods that automate scleral spur detection can facilitate qualitative and quantitative assessments of AS-OCT images.
开发一种可在眼前节光学相干断层扫描（AS-OCT）图像中检测巩膜突的深度神经网络。
参与加利福尼亚州洛杉矶的中美眼研究的参与者接受了全面的眼科检查，包括使用 Tomey CASIA SS-1000 进行 AS-OCT 成像。一名人类专家评分员在所有图像中提供了巩膜突位置的参考标签。基于 ResNet-18 架构的卷积神经网络（CNN）被开发用于检测每张图像中的巩膜突。通过计算参考和 CNN 预测的巩膜突位置之间的笛卡尔坐标差异，评估 CNN 模型的预测错误，以评估 CNN 模型的性能。将预测错误与参考评分员检测巩膜突位置的组内变异性进行比较。
使用 17704 张图像的训练数据集开发了 CNN，并使用 921 张图像的独立数据集对其进行了测试。CNN 模型的平均绝对预测误差为 X 坐标 49.27±42.07µm，Y 坐标 47.73±39.70µm。X 坐标的平均绝对组内变异性为 52.31±47.75µm，Y 坐标为 45.88±45.06µm。对于 X 坐标（ = 0.609）和 Y 坐标（ = 0.378），CNN 的预测误差分布与参考评分员的组内变异性相似。CNN 的平均绝对预测误差为 73.08±52.06µm，平均绝对组内变异性为 73.92±60.72µm。
深度神经网络可以在 AS-OCT 图像上检测巩膜突，其性能与人类专家评分员相似。
蒋锐
