---
title: CPathPatchFeature
canonical_url: "https://www.modelscope.cn/datasets/HENGFANG/CPathPatchFeature"
md_url: "https://www.modelscope.cn/datasets/HENGFANG/CPathPatchFeature.md"
repository: HENGFANG/CPathPatchFeature
chinese_name: "计算病理学图块特征"
last_updated: 2026-06-27
license: "Apache License 2.0"
storage_size: "613 GB"
downloads: 44875
stars: 1
---

# CPathPatchFeature

> CPathPatchFeature - HENGFANG 在 ModelScope 开源的数据集。Revisiting Data Challenges of Computational Pathology: A Pack-based Multiple Instance Learning Framework

HENGFANG/CPathPatchFeature 是 ModelScope 魔搭社区上的数据集，存储大小 613 GB，采用 Apache License 2.0 许可。

- **Repository**: HENGFANG/CPathPatchFeature
- **License**: Apache License 2.0
- **Storage size**: 613 GB
- **Downloads**: 44875
- **Stars**: 1
- **Last updated**: 2026-06-27

Source: https://www.modelscope.cn/datasets/HENGFANG/CPathPatchFeature

---

<h1 align="center"> Revisiting Data Challenges of Computational Pathology: A Pack-based Multiple Instance Learning Framework </h1>

<p align="center">
<a href="https://arxiv.org/abs/2509.20923"><img src="https://img.shields.io/badge/arXiv-Paper-b31b1b?logo=Arxiv"></a>
<a href=""><img src="https://img.shields.io/badge/HuggingFace-Data-ffd21e?logo=huggingface"></a>
<a href="https://pan.baidu.com/s/1OuiIP3sB68IGZeId4s4K7Q?pwd=ujtq"><img src="https://img.shields.io/badge/百度网盘-数据-06a7ff?logo=baidu"></a>
<!-- <a href=""><img src="https://img.shields.io/badge/知乎-中文讲解-1772f6?logo=zhihu"></a> -->
</p>

<h5 align="center"><em>Wenhao Tang<sup>*</sup>, Heng Fang<sup>*</sup>, Ge Wu, Xiang Li<sup>†</sup>, Ming-Ming Cheng<sup>†</sup> </em></h5>

# 项目概述

**文章地址：[[arXiv]](https://arxiv.org/abs/2509.20923) [[PDF]](https://arxiv.org/pdf/2509.20923)** 

**项目主页地址：[[GitHub]](https://github.com/FangHeng/PackMIL)**

> 注意此页面仅为数据发布页，提供给国内研究者下载实验所用数据，可能存在文字或者数据不是最新的问题，如果您在使用这些数据时遇到了任何错误，请及时联系我们进行更新（GitHub的issue或者邮箱可能是更好的选择，这个平台的留言可能看的不多）。

# 数据概述

这里提供了所有我们使用到的WSI数据特征，具体来说有：
- PANDA
	- UNI
    - CHIEF
    - GIGAP
    - R50
    - patches   # 对应特征的坐标
- BRCA
	- UNI
    - CHIEF
    - GIGAP
    - R50
    - patches   # 对应特征的坐标
- NSCLC
	- UNI
    - CHIEF
    - GIGAP
    - R50
    - patches   # 对应特征的坐标
- BLCA
    - UNI
    - CHIEF
    - GIGAP
    - R50
- CAMELYON
    - CHIEF
    - R50
    - patches   # 对应特征的坐标

> 上面每一个子文件夹都包含了对应数据集所有slide的特征，除了NSCLC和BRCA，其他的数据集都按照每个encoder的特征进行zip打包，下载后需要自行解压。BRCA和NSCLC打包后超过了ModelScope的单文件上限，所以直接按照原始结构上传了，无需进一步解压。

> 有些数据可能还没来得及整理出来，如果有需要也可以联系我们询问是否可以提供。

# 下载方式

> 这里follow了ModelScope的官方文档提供下载方式，更精细的使用方式可能需要使用者自行搜索解决。有问题也可以及时联系我们。

**SDK**

``` python
#数据集下载
from modelscope.msdatasets import MsDataset
ds =  MsDataset.load('HENGFANG/CPathPatchFeature')
#您可按需配置 subset_name、split，参照“快速使用”示例代码
```

**GIT**

确保已经正确安装了lfs：
``` bash
git lfs install
```

在此之后，直接进行克隆：
``` bash
git clone https://www.modelscope.cn/datasets/HENGFANG/CPathPatchFeature.git
```
