CheXphoto

CheXphoto是由斯坦福大学开发的一个数据集，包含超过10,000张胸透照片及其合成变换，旨在评估和提升深度学习算法在智能手机拍摄的胸透照片上的鲁棒性。该数据集主要从CheXpert数据集中抽样，通过自动和手动方式捕捉不同环境下的照片，并生成模拟摄影效果的合成变换。CheXphoto特别适用于解决资源有限地区通过智能手机进行远程医疗诊断的问题，提高全球范围内胸透图像的自动化解释能力。

CheXphoto is a dataset developed by Stanford University, comprising over 10,000 chest X-ray photographs and their synthetic transformations, which aims to evaluate and improve the robustness of deep learning algorithms for chest X-ray images captured by smartphones. This dataset is primarily sampled from the CheXpert dataset, with automatic and manual approaches employed to capture photographs across diverse environments and generate synthetic transformations that simulate real photographic effects. CheXphoto is particularly suitable for addressing the challenge of smartphone-based telemedicine diagnosis in resource-limited regions, and enhancing the global automated interpretation capability of chest X-ray images.