SMILEUHURA DS6 UNetMSS3D Wodeform

🚀 SMILEUHURA_DS6_UNetMSS3D_woDeform

This is a baseline model for the vessel segmentation challenge: SMILE - UHURA (https://doi.org/10.7303/syn47164761) and is introduced in the research paper SPOCKMIP. It aims to automatically segment small vessels in 7 Tesla 3D Time - of - Flight (ToF) Magnetic Resonance Angiography (MRA) data.

🚀 Quick Start

Blood vessels in the brain supply the necessary nutrients and oxygen. Small vessel pathologies can lead to serious issues like Cerebral Small Vessel Diseases (CSVD), which is associated with neurodegeneration such as Alzheimer’s disease. With 7 Tesla MRI systems, higher - resolution images can depict very small vessels. Traditional non - Deep Learning methods for vessel segmentation often struggle with small vessels. This paper presents a deep - learning architecture for small vessel segmentation in 7T 3D ToF MRA data. The model was trained and evaluated on a small dataset of 11 subjects, with six for training, two for validation, and three for testing. A U - Net Multi - Scale Supervision - based model was trained and made equivariant to elastic deformations using deformation - aware learning to enhance generalization. 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

• Advanced Architecture: Based on UNet Multi - Scale Supervision (UNet - MSS) 3D for accurate vessel segmentation.
• Deformation - Aware Learning: Improves generalization performance by making the model equivariant to elastic deformations.
• High - Resolution Data Utilization: Capable of segmenting small vessels in 7T 3D ToF MRA data.

📦 Installation

No installation steps are provided in the original document, so this section is skipped.

💻 Usage Examples

No code examples are provided in the original document, so this section is skipped.

📚 Documentation

Model Details

This model 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

The model architecture follows the original paper but is trained on the SMILE - UHURA dataset. It serves as a baseline model for the SMILE - UHURA challenge (https://doi.org/10.7303/syn47164761) and is also used in the SPOCKMIP research.

Model Description

Model Sources

Original DS6:

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

SPOCKMIP:

• Repository: https://github.com/soumickmj/SPOCKMIP
• Preprint: https://arxiv.org/abs/2407.08655

🔧 Technical Details

The model is based on the UNet Multi - Scale Supervision architecture. It uses a small dataset of 11 subjects for training, validation, and testing. The deformation - aware learning technique is applied to make the model more robust to elastic deformations, which helps improve the generalization performance on different datasets. The model's performance is evaluated using the Dice score, achieving 80.44 ± 0.83 on the test set.

📄 License

The model is licensed under the Apache - 2.0 license.

📄 Citation

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

BibTeX:

DS6:

@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}
}

SPOCKMIP:

@article{radhakrishna2024spockmip,
  title={SPOCKMIP: Segmentation of Vessels in MRAs with Enhanced Continuity using Maximum Intensity Projection as Loss},
  author={Radhakrishna, Chethan and Chintalapati, Karthikesh Varma and Kumar, Sri Chandana Hudukula Ram and Sutrave, Raviteja and Mattern, Hendrik and Speck, Oliver and N{\"u}rnberger, Andreas and Chatterjee, Soumick},
  journal={arXiv preprint arXiv:2407.08655},
  year={2024}
}

SMILE - UHURA:

https://doi.org/10.7303/syn47164761

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.

Radhakrishna, C., Chintalapati, K. V., Kumar, S. C. H. R., Sutrave, R., Mattern, H., Speck, O., ... & Chatterjee, S. (2024). SPOCKMIP: Segmentation of Vessels in MRAs with Enhanced Continuity using Maximum Intensity Projection as Loss. arXiv preprint arXiv:2407.08655.

https://doi.org/10.7303/syn47164761