Dataset Hub

This dataset contains 425 panoramic X‑ray images with manually annotated bounding boxes and polygons, primarily for detection and segmentation tasks on dental panoramic radiographs. The annotation details include 32 teeth, restorations, dental instruments, etc.

This is a CAMUS_public dataset for image segmentation (Cardiac Multi‑structure Ultrasound Segmentation). It contains clinical examinations from 500 patients performed at the University Hospital of Saint‑Étienne, France, fully anonymized according to local ethics committee regulations. The dataset is intended for left‑ventricular ejection fraction measurement and reflects clinical variability in image quality and pathology. It is split into a training set (450 patients) and a test set (50 new patients). Raw input images are provided in .raw/.mhd format.

IMed-361M数据集是最大的公开多模态交互式医学图像分割数据集，包含640万张图像、2.734亿个掩码（每张图像56个掩码）、14种成像模式和204个分割目标。它确保了六个解剖组之间的多样性，细粒度注释，大多数掩码覆盖的图像区域小于2%，并且广泛适用，83%的图像分辨率在256×256到1024×1024之间。IMed-361M提供的掩码数量是MedTrinity-25M的14.4倍，显著超过了其他数据集的规模和掩码数量。