Open-source PULASki_ProbUNet2D_Base_VSeg - Efficient Biomedical Image Segmentation, Effective for Small Datasets Too

Pulaski ProbUNet2D Base VSeg

Developed by soumickmj

PULASki is a computationally efficient biomedical image segmentation generative tool that accurately captures expert annotation variability, particularly suitable for small datasets and class imbalance issues.

Image Segmentation

Safetensors

Open Source License:Apache-2.0 #7T MRI vessel segmentation #Probabilistic annotation learning #Expert variability modeling

Downloads 21

Release Time : 9/3/2024

Model Overview

This model is a 2D probabilistic segmentation model based on a conditional variational autoencoder architecture (Probabilistic UNet), designed for vessel segmentation in 7T Time-of-Flight MR angiography volumetric data.

Model Features

Handling expert annotation variability

Accurately captures high variability in annotations from multiple experts, improving clinical applicability of segmentation results.

Efficient learning with small datasets

Effectively learns and generates accurate segmentation results even with limited annotated data in small datasets.

Class imbalance optimization

Utilizes an improved loss function (Focal Tversky loss), particularly effective for class imbalance issues.

Uncertainty quantification

Provides probabilistic segmentation results, avoiding potential overconfidence pitfalls of traditional methods.

Model Capabilities

Medical image segmentation

Vessel structure recognition

Probabilistic prediction

Processing 7T Time-of-Flight MR angiography data

Use Cases

Medical imaging analysis

Cerebrovascular disease diagnosis

Used for early diagnosis and assessment of cerebrovascular diseases, providing precise vessel structure segmentation.

Captures variability in expert annotations, improving diagnostic accuracy.

Surgical planning

Provides precise vessel structure information for neurosurgical procedures, assisting in surgical planning.

Reduces surgical risks and improves success rates.

Medical research

Vessel morphology studies

Used for studying vessel morphology and distribution, supporting related medical research.

Provides high-precision vessel segmentation results for quantitative analysis.

🚀 PULASki_ProbUNet2D_Base_VSeg

In the field of medical imaging, numerous supervised learning-based segmentation methods encounter challenges. These include high variability in annotations from multiple experts, a shortage of labeled data, and class-imbalanced datasets. Such issues can lead to segmentations lacking the precision required for clinical analysis and may show overconfidence without proper uncertainty quantification.

Our proposed PULASki is a computationally efficient generative tool for biomedical image segmentation. It can accurately capture the variability in expert annotations, even with small datasets. Our approach utilizes an improved loss function based on statistical distances within a conditional variational autoencoder structure (Probabilistic UNet). This enhances the learning of the conditional decoder compared to the standard cross-entropy, especially for class-imbalanced problems.

🚀 Quick Start

This section provides a high - level overview of the model and its application in medical image segmentation. For more detailed usage, please refer to the following sections.

✨ Features

Addressing Annotation Challenges: Effectively handles high variability in expert annotations.
Small Dataset Adaptability: Works well with small datasets.
Improved Loss Function: Utilizes an enhanced loss function in the Probabilistic UNet structure to improve learning, especially for class - imbalanced problems.

📚 Documentation

Model Details

It was introduced in PULASki: Learning inter - rater variability using statistical distances to improve probabilistic segmentation by Soumick Chatterjee, Franziska Gaidzik, Alessandro Sciarra, Hendrik Mattern, Gábor Janiga, Oliver Speck, Andreas Nürnberger, Sahani Pathiraja.

Model Description

Property	Details
Developed by	Dr Soumick Chatterjee
Model Type	2D Probabilistic UNet (PULASki's baseline), trained (without a distribution distance loss) with Focal Tversky loss (FTL)
Task	Probabilistic vessel segmentation in 7T MRA - ToF volumes
Training Data	7T MRA - ToF volumes, details mentioned in Sec. 4.1 of https://arxiv.org/pdf/2312.15686

Model Sources

Repository: https://github.com/soumickmj/PULASki
Paper: https://arxiv.org/abs/2312.15686

📄 License

This model is released 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 the following in your publications:

BibTeX:

@article{chatterjee2023pulaski,
  title={PULASki: Learning inter - rater variability using statistical distances to improve probabilistic segmentation},
  author={Chatterjee, Soumick and Gaidzik, Franziska and Sciarra, Alessandro and Mattern, Hendrik and Janiga, G{\'a}bor and Speck, Oliver and N{\"u}rnberger, Andreas and Pathiraja, Sahani},
  journal={arXiv preprint arXiv:2312.15686},
  year={2023}
}

APA: Chatterjee, S., Gaidzik, F., Sciarra, A., Mattern, H., Janiga, G., Speck, O., Nuernberger, A., & Pathiraja, S. (2023). PULASki: Learning inter - rater variability using statistical distances to improve probabilistic segmentation. arXiv preprint arXiv:2312.15686.

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