CrossMoDA

该数据集名为CrossMoDA，包含了来自伦敦和蒂尔堡两个不同机构的数据，其中包含了前庭神经鞘瘤和耳蜗的扫描分割。该数据集涵盖了不同分辨率和协议的扫描，为了统一性，所有图像都经过了重采样处理。在规模上，该数据集包含了210个带有注释的ceT1扫描和210个不带注释的hrT2扫描用于训练，以及来自两个地点的各105个ceT1和hrT2扫描。任务是对磁共振图像中的前庭神经鞘瘤和耳蜗进行无监督分割。

The dataset is named CrossMoDA. It contains data from two distinct institutions, London and Tilburg, including scanned segmentations of vestibular schwannomas and cochleae. The dataset covers scans with varying resolutions and imaging protocols. To ensure uniformity, all images have been resampled. In terms of scale, the dataset includes 210 annotated ceT1 scans and 210 unannotated hrT2 scans for training, as well as 105 ceT1 scans and 105 hrT2 scans from each of the two sites respectively. The task is unsupervised segmentation of vestibular schwannomas and cochleae from magnetic resonance images.