CheXphoto

该数据集包含了从原始CheXpert数据集中选取的3000名独特患者的10507张X光片，这些X光片既有自然的也有合成的转换。每张标记的X光片都针对14个标签进行了注释，重点集中在5类疾病上，分别是肺不张、心脏肥大、实变、水肿和胸腔积液。数据规模包括10507张训练图像和244张验证图像，任务是对智能手机拍摄的胸部X光片进行多标签疾病识别。

This dataset comprises 10,507 X-ray images from 3,000 unique patients selected from the original CheXpert dataset, including both naturally acquired and synthetically transformed radiographs. Each annotated X-ray image is assigned 14 labels, with a primary focus on five disease categories: atelectasis, cardiomegaly, consolidation, edema, and pleural effusion. The dataset includes 10,507 training images and 244 validation images, and the task is multi-label disease recognition for chest X-rays captured by smartphones.