Data for: Improved accuracy in OCT diagnosis of rare retinal disease using few-shot learning with generative adversarial networks

The retinal OCT images ofr rare diseases were extracted by using the Google image and Google dataset search that included English keywords including central serous chorioretinopathy (CSC), macular telangiectasia (MacTel), macular hole (MH), Stargardt disease, retinitis pigmentosa (RP). These rare diseases were selected according to a previous review article about OCT diagnosis. The images possessing rare diseases were manually classified by two board-certified ophthalmologists, and ambiguous images were removed to clarify the image domains. Additional file "Segmentation_manual.zip" offers manually segmentedOCT images for pathologic lesions.

本数据集的罕见病相关视网膜光学相干断层成像（Optical Coherence Tomography，OCT）图像，通过谷歌图片搜索及谷歌数据集搜索平台获取，检索所用英文关键词包括中心性浆液性脉络膜视网膜病变（central serous chorioretinopathy，CSC）、黄斑毛细血管扩张症（macular telangiectasia，MacTel）、黄斑裂孔（macular hole，MH）、斯塔加特病（Stargardt disease）及视网膜色素变性（retinitis pigmentosa，RP）。上述罕见病的筛选依据为既往一篇关于OCT诊断的综述文献。两名经委员会认证的眼科医师对包含罕见病特征的图像进行人工分类，并移除模棱两可的图像以明确数据集的图像域范围。附加文件"Segmentation_manual.zip"提供了针对病理病灶的人工分割OCT图像。