Prostate dataset

Samples

It is a multi-center dataset with T2 weighted modality ($C_1$, patients (n) =160; $C_2$, n=48; $C_3$, n=32; $C_4$, n=35), with a total of 279 patients. For each specific patient, we follow the annotation files (xxx.nii.gz) to obtain the masks, then extract the images based on the masks. For example, if the annotation file said this patient has three slides that contains the abnormal regions (i.e., cancers), then we will extract those three slides. The acquisition equipment varies from center to center, i.e. $C_1$ has MAGNETOM Skyra, $C_2$ has UMR 780, $C_3$ has Discovery MR750w 3.0T, and $C_4$ has MAGNETOM Verio. It has two classes, namely NMIBC (Non-muscle-invasive Bladder Cancer) and MIBC (Muscle-invasive Bladder Cancer).

Potential application areas

Because this is a multi-center dataset, so test federated learning techniques on this dataset could be of great value. Furthermore, implementing transfer learning or domain adaptation, domain generalization are also suitable.

Citations

If you found this dataset is useful, please consider to cite their work as follows.
@article{cao2024multicenter,
title={A multicenter bladder cancer MRI dataset and baseline evaluation of federated learning in clinical application},
author={Cao, Kangyang and Zou, Yujian and Zhang, Chang and Zhang, Weijing and Zhang, Jie and Wang, Guojie and Zhang, Chu and Lyu, Jiegeng and Sun, Yue and Zhang, Hongyuan and others},
journal={Scientific Data},
volume={11},
number={1},
pages={1147},
year={2024},
publisher={Nature Publishing Group UK London}
}