File size: 3,149 Bytes
a6f1dae 139da5b a6f1dae 139da5b a6f1dae 139da5b a6f1dae 7a8e02e a6f1dae 139da5b a6f1dae 139da5b a6f1dae 7a8e02e a6f1dae 7a8e02e a6f1dae 139da5b a6f1dae 7a8e02e 139da5b | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 | ---
datasets:
- AnonRes/OpenMind
license: cc-by-4.0
pipeline_tag: image-segmentation
tags:
- medical
---
# Primus-M (SimCLR Pre-trained) — OpenMind Benchmark
> **Model from the paper**: [Primus: Enforcing Attention Usage for 3D Medical Image Segmentation](https://huggingface.co/papers/2503.01835)
> **Benchmark paper**: [An OpenMind for 3D medical vision self-supervised learning](https://arxiv.org/abs/2412.17041)
> **Pre-training codebase used to create checkpoint**: [MIC-DKFZ/nnssl](https://github.com/MIC-DKFZ/nnssl)
> **Dataset**: [AnonRes/OpenMind](https://huggingface.co/datasets/AnonRes/OpenMind)
> **Downstream (segmentation) fine-tuning**: [TaWald/nnUNet](https://github.com/TaWald/nnUNet)
---
## Overview
This repository hosts a pre-trained checkpoint from the **OpenMind** benchmark using the **Primus-M** backbone. Primus is a Transformer-centric segmentation architecture designed specifically to maximize the effectiveness of attention mechanisms in 3D medical imaging, surpassing hybrid architectures and competing with state-of-the-art CNNs.
Each model in this benchmark was pre-trained using a particular self-supervised learning (SSL) method on the [OpenMind Dataset](https://huggingface.co/datasets/AnonRes/OpenMind), a large-scale, standardized collection of public brain MRI datasets.
**Note**: These models are primarily intended for downstream fine-tuning. We recommend using the adaptation frameworks for segmentation and classification provided in the [nnU-Net adaptation repository](https://github.com/TaWald/nnUNet).
*While manual download is possible, we recommend using the auto-download feature of the fine-tuning repository by providing the repository URL on Hugging Face instead of a local checkpoint path.*
---
## Model Variants
In the OpenMind benchmark, SSL checkpoints are released for two backbone architectures:
- **ResEnc-L**: A CNN-based encoder [[a](https://arxiv.org/abs/2410.23132), [b](https://arxiv.org/abs/2404.09556)]
- **Primus-M**: A transformer-based encoder introduced in the [Primus paper](https://huggingface.co/papers/2503.01835).
Each encoder has been pre-trained using various SSL techniques such as SimCLR, Volume Contrastive (VoCo), VolumeFusion (VF), and Masked Autoencoders (MAE). This specific checkpoint uses **SimCLR**.
## Citation
If you use this model, please cite:
```bibtex
@article{wald2025primus,
title={Primus: Enforcing Attention Usage for 3D Medical Image Segmentation},
author={Wald, Tassilo and Roy, Saikat and Isensee, Fabian and Ulrich, Constantin and Ziegler, Sebastian and Trofimova, Dasha and Stock, Raphael and Baumgartner, Michael and Köhler, Gregor and Maier-Hein, Klaus},
journal={arXiv preprint arXiv:2503.01835},
year={2025}
}
@article{wald2024openmind,
title={An OpenMind for 3D medical vision self-supervised learning},
author={Wald, Tassilo and Ulrich, Constantin and Suprijadi, J. and Ziegler, Sebastian and Nohel, M. and Peretzke, R. and others},
journal={arXiv preprint arXiv:2412.17041},
year={2024}
}
``` |