Viterbi Family Department of Ophthalmology, University of California San Diego, La Jolla, California, USA.
Ophthalmology and Vision Science, University of Louisville, Louisville, Kentucky, USA.
Br J Ophthalmol. 2024 Nov 22;108(12):1688-1693. doi: 10.1136/bjo-2023-324528.
BACKGROUND/AIMS: To design a deep learning (DL) model for the detection of glaucoma progression with a longitudinal series of macular optical coherence tomography angiography (OCTA) images.
202 eyes of 134 patients with open-angle glaucoma with ≥4 OCTA visits were followed for an average of 3.5 years. Glaucoma progression was defined as having a statistically significant negative 24-2 visual field (VF) mean deviation (MD) rate. The baseline and final macular OCTA images were aligned according to centre of fovea avascular zone automatically, by checking the highest value of correlation between the two images. A customised convolutional neural network (CNN) was designed for classification. A comparison of the CNN to logistic regression model for whole image vessel density (wiVD) loss on detection of glaucoma progression was performed. The performance of the model was defined based on the confusion matrix of the validation dataset and the area under receiver operating characteristics (AUC).
The average (95% CI) baseline VF MD was -3.4 (-4.1 to -2.7) dB. 28 (14%) eyes demonstrated glaucoma progression. The AUC (95% CI) of the DL model for the detection of glaucoma progression was 0.81 (0.59 to 0.93). The sensitivity, specificity and accuracy (95% CI) of DL model were 67% (34% to 78%), 83% (42% to 97%) and 80% (52% to 95%), respectively. The AUC (95% CI) for the detection of glaucoma progression based on the logistic regression model was lower than the DL model (0.69 (0.50 to 0.88)).
The optimised DL model detected glaucoma progression based on longitudinal macular OCTA images showed good performance. With external validation, it could enhance detection of glaucoma progression.
NCT00221897.
背景/目的：设计一种基于纵向系列黄斑光学相干断层扫描血管造影（OCTA）图像的深度学习（DL）模型，用于检测青光眼进展。
对 134 例开角型青光眼患者的 202 只眼进行了≥4 次 OCTA 随访，平均随访时间为 3.5 年。青光眼进展定义为视野（VF）平均缺损（MD）率有统计学意义的负 24-2。根据黄斑 OCTA 图像中中心无血管区（FAZ）自动对齐基线和最终图像，通过检查两幅图像之间相关性的最大值来实现。设计了一个定制的卷积神经网络（CNN）用于分类。比较了 CNN 与逻辑回归模型对整个图像血管密度（wiVD）损失检测青光眼进展的性能。根据验证数据集的混淆矩阵和接收者操作特征（ROC）曲线下面积（AUC）来定义模型的性能。
平均（95%CI）基线 VF MD 为-3.4（-4.1 至-2.7）dB。28 只（14%）眼出现青光眼进展。DL 模型检测青光眼进展的 AUC（95%CI）为 0.81（0.59 至 0.93）。DL 模型检测青光眼进展的敏感性、特异性和准确性（95%CI）分别为 67%（34%至 78%）、83%（42%至 97%）和 80%（52%至 95%）。基于逻辑回归模型检测青光眼进展的 AUC（95%CI）低于 DL 模型（0.69（0.50 至 0.88））。
基于纵向黄斑 OCTA 图像优化的 DL 模型检测青光眼进展的性能良好。经过外部验证，它可以提高青光眼进展的检测能力。
NCT00221897。
