Viterbi Family Department of Ophthalmology, Shiley Eye Institute, Hamilton Glaucoma Center.
Department of Ophthalmology, University Medical Center of the Johannes Gutenberg University Mainz.
J Glaucoma. 2023 Oct 1;32(10):841-847. doi: 10.1097/IJG.0000000000002267. Epub 2023 Jul 19.
An optical coherence tomography (OCT)-based multimodal deep learning (DL) classification model, including texture information, is introduced that outperforms single-modal models and multimodal models without texture information for glaucoma diagnosis in eyes with and without high myopia.
BACKGROUND/AIMS: To evaluate the diagnostic accuracy of a multimodal DL classifier using wide OCT optic nerve head cube scans in eyes with and without axial high myopia.
Three hundred seventy-one primary open angle glaucoma (POAG) eyes and 86 healthy eyes, all without axial high myopia [axial length (AL) ≤ 26 mm] and 92 POAG eyes and 44 healthy eyes, all with axial high myopia (AL > 26 mm) were included. The multimodal DL classifier combined features of 3 individual VGG-16 models: (1) texture-based en face image, (2) retinal nerve fiber layer (RNFL) thickness map image, and (3) confocal scanning laser ophthalmoscope (cSLO) image. Age, AL, and disc area adjusted area under the receiver operating curves were used to compare model accuracy.
Adjusted area under the receiver operating curve for the multimodal DL model was 0.91 (95% CI = 0.87, 0.95). This value was significantly higher than the values of individual models [0.83 (0.79, 0.86) for texture-based en face image; 0.84 (0.81, 0.87) for RNFL thickness map; and 0.68 (0.61, 0.74) for cSLO image; all P ≤ 0.05]. Using only highly myopic eyes, the multimodal DL model showed significantly higher diagnostic accuracy [0.89 (0.86, 0.92)] compared with texture en face image [0.83 (0.78, 0.85)], RNFL [0.85 (0.81, 0.86)] and cSLO image models [0.69 (0.63, 0.76)] (all P ≤ 0.05).
Combining OCT-based RNFL thickness maps with texture-based en face images showed a better ability to discriminate between healthy and POAG than thickness maps alone, particularly in high axial myopic eyes.
一种基于光学相干断层扫描（OCT）的多模态深度学习（DL）分类模型，包括纹理信息，用于诊断有和无高度近视的眼中的青光眼，其性能优于单模态模型和没有纹理信息的多模态模型。
背景/目的：评估一种使用宽 OCT 视盘立方体扫描的多模态 DL 分类器在无轴向高度近视[眼轴长度（AL）≤26mm]和有轴向高度近视（AL>26mm）的眼中诊断原发性开角型青光眼（POAG）的准确性。
纳入 371 只原发性开角型青光眼（POAG）眼和 86 只健康眼，均无轴向高度近视[眼轴长度（AL）≤26mm]和 92 只 POAG 眼和 44 只健康眼，均有轴向高度近视（AL>26mm）。多模态 DL 分类器结合了 3 个单独的 VGG-16 模型的特征：（1）基于纹理的眼底图像，（2）视网膜神经纤维层（RNFL）厚度图图像，和（3）共焦扫描激光检眼镜（cSLO）图像。使用年龄、AL 和视盘面积调整的接收者操作特征曲线下面积来比较模型准确性。
多模态 DL 模型的调整后接收者操作特征曲线下面积为 0.91（95%置信区间=0.87，0.95）。这一值明显高于各个单模型的值[基于纹理的眼底图像为 0.83（0.79，0.86）；RNFL 厚度图为 0.84（0.81，0.87）；和 cSLO 图像为 0.68（0.61，0.74）；所有 P 值均≤0.05]。仅使用高度近视眼，多模态 DL 模型的诊断准确性明显更高[0.89（0.86，0.92）]，明显优于纹理眼底图像[0.83（0.78，0.85）]、RNFL[0.85（0.81，0.86）]和 cSLO 图像模型[0.69（0.63，0.76）]（所有 P 值均≤0.05）。
将基于 OCT 的 RNFL 厚度图与基于纹理的眼底图像相结合，比单独使用厚度图能够更好地区分健康眼和 POAG，特别是在高度轴性近视眼中。
