ViMed-PET-CT
收藏资源简介:
ViMed-PET-CT数据集是一个经过优化压缩的医学影像数据集,源自ViMed-PET,将原有的.npy和分块zip文件转换为.npz格式,并提供了更好的标注和指南。数据集包含2017、2018、2019和2023年的数据,每位患者的数据包括基本元数据(性别、身高、体重)、CT扫描、PET扫描以及生成的报告。此外,数据集还包含一个独立的`medical_test_set/`文件夹。CT扫描的尺寸为(313, 512, 512),PET扫描的尺寸为(313, 256, 256)。元数据文件`metadata.csv`包含了性别、身高、体重、年份、PET和CT的路径以及报告路径等信息。该数据集适用于图像到文本、文本生成和图像文本到文本等任务,规模在1K到10K之间。数据集采用cc-by-4.0许可,并引用于一项关于医学数据视觉语言基础模型的研究。
ViMed-PET-CT dataset is an optimized and compressed medical imaging dataset derived from ViMed-PET. It converts the original .npy and chunked zip files into .npz format, and provides improved annotations and guidelines. The dataset contains data from the years 2017, 2018, 2019 and 2023. For each patient, the data includes basic metadata (gender, height, weight), CT scans, PET scans and generated reports. Additionally, the dataset contains a standalone `medical_test_set/` folder. The dimensions of CT scans are (313, 512, 512), while those of PET scans are (313, 256, 256). The metadata file `metadata.csv` contains information such as gender, height, weight, year, paths of PET and CT scans, and report paths. This dataset is applicable to tasks including image-to-text, text generation and image-text-to-text, with a scale ranging from 1K to 10K. The dataset is licensed under CC BY 4.0, and has been cited in a study on visual-language foundation models for medical data.




