README.md

---
license: cc-by-nc-4.0
datasets:
- ODELIA-AI/ODELIA-Challenge-2025
language:
- en
metrics:
- roc_auc
pipeline_tag: image-classification
tags:
- breast
- cancer
- odelia
extra_gated_prompt: >-
  ### 🛡️ Model Usage Agreement

  By accessing or using this model (the “Model”), you acknowledge and agree to the following terms and conditions:

  #### 1. Research-Only Use

  The Model is provided strictly for non-commercial, academic, and research purposes. It must not be used for clinical decision-making, diagnosis, treatment, or any other application involving real patients or clinical care.

  #### 2. No Clinical or Commercial Deployment

  The Model is **not approved for clinical use** or any commercial application. Any deployment in healthcare settings or use for patient-related decision support is expressly prohibited.

  #### 3. Redistribution and Modification

  You may not copy, distribute, sublicense, or otherwise share the Model or any derivative works without prior written permission from the model authors or the ODELIA consortium.

  #### 4. Privacy and Ethics Compliance

  You must not attempt to identify, re-identify, or deanonymize any individual whose data may have contributed to the training or evaluation of the Model.

  #### 5. Attribution Requirement

  Any publication, presentation, or derivative work that uses or references this Model must include clear attribution to the **ODELIA consortium**, along with any citations specified in the accompanying documentation.

  #### 6. Responsibility and Verification

  You are solely responsible for verifying and validating the Model’s outputs and ensuring they are appropriate for your research context. The Model and its outputs are provided “as is,” without warranties of any kind.

  #### 7. Inclusion of Third-Party Components

  This Model incorporates or is derived from **DINOv3**, developed by **Meta Platforms**.
  Use of the Model is therefore also subject to the **DINOv3 License Agreement**.
  By using this Model, you agree to comply with both:

  * This Model Usage Agreement, **and**
  * The [DINOv3 License Terms](https://github.com/facebookresearch/dinov3).
---

# ODELIA Classification Baseline Model
For a comprehensive description of the model and its intended use, please refer to our paper: [Read the paper](https://arxiv.org/abs/2506.00474)

## Setup

To run the code, we recommend creating a Python virtual environment.

### Using venv

```bash
# Create a virtual environment
python -m venv venv

# Activate the environment
# On Linux/Mac:
source venv/bin/activate
# On Windows:
# venv\Scripts\activate

# Install dependencies
pip install torch torchvision numpy huggingface_hub torchio matplotlib transformers einops x_transformers
```

### Using Conda

```bash
# Create a conda environment
conda create -n odelia_hf python=3.10
conda activate odelia_hf

# Install dependencies
pip install torch torchvision numpy huggingface_hub torchio matplotlib transformers einops x_transformers
```

## Get Probabilities and Attention

To use this model, first download the required files from this repository:

```python
from huggingface_hub import hf_hub_download

# Download model files to local directory
hf_hub_download(repo_id="ODELIA-AI/MST", filename="models.py", local_dir="./")
hf_hub_download(repo_id="ODELIA-AI/MST", filename="predict_attention.py", local_dir="./")
```

Then execute `predict_attention.py --path_img path/to/Sub_1.nii.gz` to get probabilities and attention maps.