ResEncL-OpenMind-VoCo Open-Source Model - Empowering Self-Supervised Learning Research on 3D Medical Imaging Data

Resencl OpenMind VoCo

Developed by AnonRes

The first comprehensive benchmark study model for self-supervised learning on 3D medical imaging data

3D Vision #3D Medical Imaging #Self-supervised Pretraining #Multimodal SSL

Downloads 16

Release Time : 5/6/2025

Model Overview

This model provides a series of 3D medical imaging pretraining checkpoints based on different self-supervised learning methods, primarily for medical image analysis tasks, especially brain MRI data processing.

Model Features

Diverse Self-supervised Learning Methods

Provides pretrained models with 8 different self-supervised learning techniques, including VoCo, VF, MG, MAE, etc.

Dual Architecture Support

Offers both CNN-based and Transformer-based backbone architecture options

Standardized Medical Dataset

Trained on the OpenMind dataset, a large-scale, standardized collection of public brain MRI datasets

Downstream Task Adaptation

Specifically designed for downstream tasks like medical image segmentation, with fine-tuning framework support

Model Capabilities

3D Medical Imaging Feature Extraction

Brain MRI Analysis

Medical Image Segmentation

Self-supervised Learning Pretraining

Use Cases

Medical Imaging Analysis

Brain MRI Segmentation

Used for tissue structure segmentation in brain MRI images

Performs excellently on standardized datasets

Medical Imaging Pretraining

Serves as a pretraining foundation model for medical imaging analysis tasks

Significantly improves performance on downstream tasks

🚀 OpenMind Benchmark 3D SSL Models

This repository provides pre - trained checkpoints for 3D medical vision self - supervised learning, aiming to facilitate research and development in the field of 3D medical imaging.

🚀 Quick Start

The models in this repository are from the OpenMind benchmark. Here are some key links:

Model from the paper: An OpenMind for 3D medical vision self - supervised learning
Pre - training codebase used to create checkpoint: [MIC - DKFZ/nnssl](https://github.com/MIC - DKFZ/nnssl)
Dataset: AnonRes/OpenMind
Downstream (segmentation) fine - tuning: TaWald/nnUNet

OpenMind

✨ Features

Overview

This repository hosts pre - trained checkpoints from the OpenMind benchmark:
📄 An OpenMind for 3D medical vision self - supervised learning (Wald, T., Ulrich, C., Suprijadi, J., Ziegler, S., Nohel, M., Peretzke, R., ... & Maier - Hein, K. H. (2024).)
(arXiv:2412.17041) — the first extensive benchmark study for self - supervised learning (SSL) on 3D medical imaging data.

Each model was pre - trained using a particular SSL method on the OpenMind Dataset, a large - scale, standardized collection of public brain MRI datasets.

⚠️ Important Note

These models are not recommended to be used as - is for feature extraction. Instead we recommend using the downstream fine - tuning frameworks for segmentation and classification adaptation, available in the adaptation repository.

💡 Usage Tip

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

We release SSL checkpoints for two backbone architectures:

ResEnc - L: A CNN - based encoder [a, b]
Primus - M: A transformer - based encoder [Primus paper]

Each encoder has been pre - trained using one of the following SSL techniques:

Property	Details
Model Type	We have models based on two backbone architectures: ResEnc - L (CNN - based) and Primus - M (transformer - based).
Training Data	The models are pre - trained on the OpenMind Dataset, a large - scale, standardized collection of public brain MRI datasets.

Method	Description
Volume Contrastive (VoCo)	Contrastive pretraining method for 3D volumes
VolumeFusion (VF)	Spatial volume fusion - based segmentation SSL method
Models Genesis (MG)	Reconstruction and denoising based pretraining method
Masked Autoencoders (MAE)	Default reconstruction based pretraining method
Spark 3D (S3D)	Sparse reconstruction based pretraining mehtod (CNN only)
SimMIM	Simple masked reconstruction based pretraining method (TR only)
SwinUNETR SSL	Rotation, Contrastive and Reconstruction based pre - training method.
SimCLR	Transfer of 2D Contrastive learning baseline method to 3D

📄 License

The license for this project is cc - by - 4.0.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご