OCT Dataset

该数据集包括经标注的光学相干断层扫描（OCT）图像，涵盖四种类型：脉络膜新生血管（CNV）、糖尿病性黄斑水肿（DME）、视网膜下脉络膜积液（Drusen）和正常图像。总计样本量为83,484个。具体来说，训练数据集包含四个类别的图像，分别为{CNV: 26,318个，DME: 8,616个，Drusen: 11,350个，正常: 37,205个}，而测试集则包含968个图像，同样分为这四个类别。该数据集的总规模为83,484个样本，任务为医学图像分类。

This dataset contains annotated Optical Coherence Tomography (OCT) images, covering four categories: Choroidal Neovascularization (CNV), Diabetic Macular Edema (DME), Subretinal Choroidal Effusion (Drusen), and normal OCT images. The total number of samples is 83,484. Specifically, the training dataset includes images from these four categories, with the following counts: CNV: 26,318, DME: 8,616, Drusen: 11,350, and normal: 37,205. The test set consists of 968 images, which are also divided into these four categories. The overall scale of this dataset is 83,484 samples, and the corresponding task is medical image classification.

搜集汇总

背景与挑战

背景概述

该数据集是一个关于视网膜光学相干断层扫描(OCT)图像的数据集，主要用于通过监督和半监督学习技术进行疾病识别。数据集包含训练脚本、超参数优化配置以及相关实验结果。