DS6_UNetMSS3D_wDeform Open-source Model - Accurately Segment Small Brain Blood Vessels in 7T MRA

DS6 UNetMSS3D Wdeform

Developed by soumickmj

A 3D deep learning model based on UNet multi-scale supervision for small vessel segmentation in 7T MRA-ToF data, utilizing deformation-aware learning to enhance generalization performance.

Image Segmentation

Safetensors

Open Source License:Apache-2.0 #7T small vessel segmentation #Deformation-aware learning #3D UNet multi-scale supervision

Downloads 187

Release Time : 9/1/2024

Model Overview

This model is specifically designed for 7 Tesla 3D Time-of-Flight Magnetic Resonance Angiography data, capable of automatically segmenting tiny brain vessels, particularly targeting small vessel structures that are difficult to identify with traditional methods.

Model Features

Small vessel segmentation optimization

Specifically optimized for tiny brain vessel structures that are difficult to detect with traditional methods, significantly improving small vessel recognition rates.

Deformation-aware learning

Utilizes self-supervised DS6 deformation-aware learning technology, making the model equivariant to elastic deformations and improving generalization performance by 18.98%.

Few-shot learning

Achieves good performance with minimal training data (6 semi-automatically segmented samples), suitable for scenarios with scarce medical imaging data.

Multi-scale supervision

Based on the UNet-MSS architecture's multi-scale supervision mechanism, enhancing the extraction capability of vessel features at different scales.

Model Capabilities

3D medical image segmentation

Small vessel detection

Magnetic resonance image analysis

Brain vascular network reconstruction

Use Cases

Medical research

Cerebral small vessel disease research

Used for quantitative analysis of microvascular lesions in CSVD-related studies

Dice score reached 80.44±0.83

Neurodegenerative disease research

Assists in studying the relationship between neurodegenerative diseases like Alzheimer's and vascular abnormalities

Clinical assistance

Preoperative vascular assessment

Provides high-precision 3D visualization of vascular networks for neurosurgical procedures

🚀 DS6_UNetMSS3D_wDeform

This project proposes a deep - learning architecture for automatically segmenting small vessels in 7 Tesla 3D Time - of - Flight (ToF) Magnetic Resonance Angiography (MRA) data, aiming to address the challenges in small vessel segmentation.

🚀 Quick Start

Blood vessels in the brain supply the necessary nutrients and oxygen. Small vessel pathology can lead to severe issues like Cerebral Small Vessel Diseases (CSVD), which is also associated with neurodegeneration such as Alzheimer’s disease. With 7 Tesla MRI systems, higher - resolution images can depict very small vessels. Non - Deep Learning - based vessel segmentation methods often struggle with small vessels. This paper presents a deep - learning architecture to automatically segment small vessels in 7 Tesla 3D ToF MRA data. The algorithm was trained and evaluated on a small dataset of 11 subjects. A U - Net Multi - Scale Supervision - based deep - learning model was trained using deformation - aware learning to enhance generalization performance. The proposed technique achieved a Dice score of 80.44 ± 0.83 on the test set and showed an 18.98% improvement compared to a manually segmented region.

✨ Features

Automatically segment small vessels in 7 Tesla 3D ToF MRA data.
Use deformation - aware learning to improve generalization performance.
Achieve a high Dice score in small vessel segmentation.

📚 Documentation

Model Details

It was introduced in DS6, Deformation - Aware Semi - Supervised Learning: Application to Small Vessel Segmentation with Noisy Training Data by Soumick Chatterjee, Kartik Prabhu, Mahantesh Pattadkal, Gerda Bortsova, Chompunuch Sarasaen, Florian Dubost, Hendrik Mattern, Marleen de Bruijne, Oliver Speck, Andreas Nürnberger. ArXiv preprint

Model Description

Property	Details
Model Type	UNet Multi - scale Supervision (UNet - MSS) 3D
Task	Vessel segmentation in 7T MRA - ToF volumes
Training Dataset	In - house 7T ToF - MRAs, details can be found in Section 4 of the published paper.
Training Type	Trained with deformation - aware learning (Proposed DS6 method)

Model Sources

Repository: https://github.com/soumickmj/DS6
Paper: https://doi.org/10.3390/jimaging8100259
Preprint: https://arxiv.org/abs/2006.10802

Citation

If you use this approach in your research or use codes from this repository or these weights, please cite the following in your publications:

BibTeX:

@article{chatterjee2022ds6,
  title={Ds6, deformation - aware semi - supervised learning: Application to small vessel segmentation with noisy training data},
  author={Chatterjee, Soumick and Prabhu, Kartik and Pattadkal, Mahantesh and Bortsova, Gerda and Sarasaen, Chompunuch and Dubost, Florian and Mattern, Hendrik and de Bruijne, Marleen and Speck, Oliver and N{\"u}rnberger, Andreas},
  journal={Journal of Imaging},
  volume={8},
  number={10},
  pages={259},
  year={2022},
  publisher={MDPI}
}

APA: Chatterjee, S., Prabhu, K., Pattadkal, M., Bortsova, G., Sarasaen, C., Dubost, F., ... & Nürnberger, A. (2022). Ds6, deformation - aware semi - supervised learning: Application to small vessel segmentation with noisy training data. Journal of Imaging, 8(10), 259.

📄 License

This project is licensed under the Apache - 2.0 license.

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