Jain Rishabh, Yoo Tae Keun, Ryu Ik Hee, Song Joanna, Kolte Nitin, Nariani Ashiyana
Department of Biomedical Engineering, Duke University, Durham, NC, USA.
Department of Refractive Surgery, B&VIIT Eye Center, B2 GT Tower, 1317-23 Seocho-Dong, Seocho-Gu, Seoul, Republic of Korea.
Ophthalmol Ther. 2024 Jan;13(1):305-319. doi: 10.1007/s40123-023-00842-6. Epub 2023 Nov 13.
The mismatch between training and testing data distribution causes significant degradation in the deep learning model performance in multi-ethnic scenarios. To reduce the performance differences between ethnic groups and image domains, we built a deep transfer learning model with adaptation training to predict uncorrected refractive errors using posterior segment optical coherence tomography (OCT) images of the macula and optic nerve.
Observational, cross-sectional, multicenter study design. We pre-trained a deep learning model on OCT images from the B&VIIT Eye Center (Seoul, South Korea) (N = 2602 eyes of 1301 patients). OCT images from Poona Eye Care (Pune, India) were chronologically sorted into adaptation training data (N = 60 eyes of 30 patients) for transfer learning and test data (N = 142 eyes of 71 patients) for validation. Deep learning models were trained to predict spherical equivalent (SE) and mean keratometry (K) values via transfer learning for domain adaptation.
Both adaptation models for SE and K were significantly better than those without adaptation (P < 0.001). In myopia/hyperopia classification, the model trained on circular optic disc OCT images yielded the best performance (accuracy = 74.7%). It also performed best in estimating SE with the lowest mean absolute error (MAE) of 1.58 D. For classifying the degree of corneal curvature, the optic nerve vertical algorithm performed best (accuracy = 65.7%). The optic nerve horizontal model achieved the lowest MAE (1.85 D) when predicting the K value. Saliency maps frequently highlighted the retinal nerve fiber layers.
Adaptation training via transfer learning is an effective technique for estimating refractive errors and K values using macular and optic nerve OCT images from ethnically heterogeneous populations. Further studies with larger sample sizes and various data sources are needed to confirm the feasibility of the proposed algorithm.
训练数据与测试数据分布不匹配会导致深度学习模型在多民族场景下的性能显著下降。为了减少不同种族群体和图像域之间的性能差异，我们构建了一个经过自适应训练的深度迁移学习模型，以使用黄斑和视神经的后段光学相干断层扫描（OCT）图像来预测未矫正的屈光不正。
观察性、横断面、多中心研究设计。我们在来自韩国首尔B&VIIT眼科中心的OCT图像上预训练了一个深度学习模型（N = 1301例患者的2602只眼）。来自印度浦那眼科护理中心的OCT图像按时间顺序分类为用于迁移学习的自适应训练数据（N = 30例患者的60只眼）和用于验证的测试数据（N = 71例患者的142只眼）。通过迁移学习进行域自适应训练深度学习模型，以预测等效球镜度（SE）和平均角膜曲率（K）值。
SE和K的两种自适应模型均显著优于未进行自适应的模型（P < 0.001）。在近视/远视分类中，在圆形视盘OCT图像上训练的模型表现最佳（准确率 = 74.7%）。在估计SE时，其平均绝对误差（MAE）最低，为1.58 D，表现也最佳。在分类角膜曲率程度时，视神经垂直算法表现最佳（准确率 = 65.7%）。在预测K值时，视神经水平模型的MAE最低（1.85 D）。显著性映射经常突出显示视网膜神经纤维层。
通过迁移学习进行自适应训练是一种使用来自不同种族人群的黄斑和视神经OCT图像估计屈光不正和K值的有效技术。需要进一步开展更大样本量和各种数据源的研究，以确认所提出算法的可行性。
