数据集获取

git clone https://github.com/ankilab/HANCOCK_MultimodalDataset.git

curl -L -o repo.zip https://github.com/ankilab/HANCOCK_MultimodalDataset/archive/refs/heads/main.zip
unzip repo.zip

源站 README 摘录（使用方式）

Multimodal head and neck cancer dataset

Website | Dataset download | Cite
This repository contains code for exploring the multimodal head and neck cancer dataset HANCOCK and for
training Machine Learning models to predict outcomes and treatment choices. We also provide strategies for
multimodal data handling, feature extraction, and generating train/test dataset splits.

Table of contents

• Setup
• Dataset
• Multimodal feature extraction
• Generating data splits
• Outcome prediction
• Reference

Setup

To set up the environment, first clone this repository to your local machine and create a directory for storing
the results:

git clone https://github.com/ankilab/HANCOCK_MultimodalDataset.git
cd HANCOCK_MultimodalDataset
mkdir results

Next, set up an Anaconda environment and install the required Python packages:

conda create -n hancock_multimodal python=3.12
conda activate hancock_multimodal
pip install -r requirements.txt

Our code was tested on Ubuntu-24.04 with an NVIDIA RTX 4060 besides the additional section
Adjuvant treatment prediction using histology images,
which was tested on Windows.
For running the code described the section
Adjuvant treatment prediction using histology images,
TensorFlow 2.16 is used (see requirements.txt). Furthermore, QuPath needs to be installed
for the analysis of histology data.

Dataset

The dataset can be explored and downloaded at our project website: https://hancock.research.fau.eu/
The dataset is structured in ZIP archives. If all archives are downloaded and unzipped, the dataset is
structured as follows:

Hancock_Dataset
|
├── StructuredData
|   ├── blood_data.json
|   ├── blood_data_reference_ranges.json
|   ├── clinical_data.json
|   └── pathological_data.json
|
├── TextData
|   ├── histories
|   ├── histories_english
|   ├── icd_codes
|   ├── ops_codes
|   ├── reports
|   ├── reports_english
|   ├── surgery_descriptions
|   └── surgery_descriptions_english
|
├── DataSplits_DataDictionaries
|   ├── DataDictionary_blood.csv
|   ├── DataDic


## 数据加载示例（图像类）

```python
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。

完整仓库：github.com/ankilab/HANCOCK_MultimodalDataset