FeTS 2022 挑战赛

数据集获取

源站：https://synapse.org/Synapse:syn28546456

标准获取流程：

1. 访问源站链接，注册/登录账号；
2. 进入数据集详情页，按页面指引申请权限或直接下载；
3. 部分平台提供 SDK，例如：
   • OpenDataLab: pip install openxlab 后 openxlab.dataset.get(...)
   • ModelScope: pip install modelscope 后 modelscope download
   • 天池: 登录后页面提供直链下载
     以源站「数据下载/使用文档」页面为准。

数据说明

###! Challenge closed ####! Thank you to everyone who joined the challenge! Congratulations to our top-performing teams! — ${image?fileName=FeTS Challenge 2022 Banner%2Epng&align=Center&scale=100&responsive=true&altText=FeTS Challenge 2022 banner} # Overview Welcome to the Federated Tumor Segmentation (FeTS) 2022 Challenge! FeTS 2022 focuses on benchmarking methods for federated learning (FL), and particularly i) weight aggregation methods for federated training, and ii) algorithmic generalizability on out-of-sample data based on federated evaluation. In line with its last instance (FeTS 2021 - the 1st FL challenge ever organized), FeTS 2022 targets the task of brain tumor segmentation and builds upon i) the centralized dataset of >8,000 clinically-acquired multi-institutional MRI scans (from the RSNA-ASNR-MICCAI BraTS 2021 challenge) with their real-world partitioning, and ii) the collaborative network of remote independent institutions included in a real-world federation. Participants are welcome to compete in either of the two challenge tasks: - Task 1 (“Federated Training”) seeks effective weight aggregation methods for the creation of a consensus model given a pre-defined segmentation algorithm for training, while also (optionally) accounting for network outages. - Task 2 (“Federated Evaluation”) seeks robust segmentation algorithms (centralized learning is allowed), evaluated during the testing phase on unseen data from various remote independent institutions of the collaborative network of the FeTS federation. Compared to the FeTS 2021 challenge, this year we have expanded both the size and diversity of the training, validation, and testing data. For those of you who previously participated in the BraTS challenge, it is worth noting that: - The FeTS Challenge will be different from BraTS, in that each patient dataset is linked to a de-identified partitioning label accordin

数据加载示例（图像类）

from PIL import Image
import glob, os

files = (glob.glob(os.path.join(path, "**", "*.png"), recursive=True)
       + glob.glob(os.path.join(path, "**", "*.jpg"), recursive=True)
       + glob.glob(os.path.join(path, "**", "*.tif"), recursive=True))
print("图像文件数:", len(files))
img = Image.open(files[0]); print("尺寸/模式:", img.size, img.mode)

# torchvision Dataset 方式:
# from torchvision import datasets
# ds = datasets.ImageFolder(path)  # 要求 子目录=类别

目录组织与标注格式以源站说明和下载后实际文件为准。