Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Improved Multi-Task Brain Tumour Segmentation with Synthetic Data Augmentation

André Ferreira, Tiago Jesus, Behrus Puladi, Jens Kleesiek, Victor Alves, Jan Egger

TL;DR

This work addresses robust brain tumor segmentation in post-treatment gliomas and radiotherapy planning for meningiomas by leveraging synthetic data augmentation with GliGANs. It integrates three architectures ( nnUNet, MedNeXt, Swin UNETR ) within an nnUNet-based pipeline, using 5-fold cross-validation and ensemble fusion to improve segmentation robustness across tasks. Key findings show competitive Dice scores on Task 1 (e.g., region-wise DSCs near 0.79–0.89) and Task 3 (DSC around 0.801, HD95 ≈ 38.26), with synthetic data contributing to reduced false negatives and improved model stability, albeit with task-specific limitations. The approach demonstrates the potential of GAN-based augmentation to enhance clinical applicability of automated segmentation tools, while highlighting areas for future improvement in tumour placement realism and synthetic-data quality assessment.

Abstract

This paper presents the winning solution of task 1 and the third-placed solution of task 3 of the BraTS challenge. The use of automated tools in clinical practice has increased due to the development of more and more sophisticated and reliable algorithms. However, achieving clinical standards and developing tools for real-life scenarios is a major challenge. To this end, BraTS has organised tasks to find the most advanced solutions for specific purposes. In this paper, we propose the use of synthetic data to train state-of-the-art frameworks in order to improve the segmentation of adult gliomas in a post-treatment scenario, and the segmentation of meningioma for radiotherapy planning. Our results suggest that the use of synthetic data leads to more robust algorithms, although the synthetic data generation pipeline is not directly suited to the meningioma task. In task 1, we achieved a DSC of 0.7900, 0.8076, 0.7760, 0.8926, 0.7874, 0.8938 and a HD95 of 35.63, 30.35, 44.58, 16.87, 38.19, 17.95 for ET, NETC, RC, SNFH, TC and WT, respectively and, in task 3, we achieved a DSC of 0.801 and HD95 of 38.26, in the testing phase. The code for these tasks is available at https://github.com/ShadowTwin41/BraTS_2023_2024_solutions.

Improved Multi-Task Brain Tumour Segmentation with Synthetic Data Augmentation

TL;DR

This work addresses robust brain tumor segmentation in post-treatment gliomas and radiotherapy planning for meningiomas by leveraging synthetic data augmentation with GliGANs. It integrates three architectures ( nnUNet, MedNeXt, Swin UNETR ) within an nnUNet-based pipeline, using 5-fold cross-validation and ensemble fusion to improve segmentation robustness across tasks. Key findings show competitive Dice scores on Task 1 (e.g., region-wise DSCs near 0.79–0.89) and Task 3 (DSC around 0.801, HD95 ≈ 38.26), with synthetic data contributing to reduced false negatives and improved model stability, albeit with task-specific limitations. The approach demonstrates the potential of GAN-based augmentation to enhance clinical applicability of automated segmentation tools, while highlighting areas for future improvement in tumour placement realism and synthetic-data quality assessment.

Abstract

This paper presents the winning solution of task 1 and the third-placed solution of task 3 of the BraTS challenge. The use of automated tools in clinical practice has increased due to the development of more and more sophisticated and reliable algorithms. However, achieving clinical standards and developing tools for real-life scenarios is a major challenge. To this end, BraTS has organised tasks to find the most advanced solutions for specific purposes. In this paper, we propose the use of synthetic data to train state-of-the-art frameworks in order to improve the segmentation of adult gliomas in a post-treatment scenario, and the segmentation of meningioma for radiotherapy planning. Our results suggest that the use of synthetic data leads to more robust algorithms, although the synthetic data generation pipeline is not directly suited to the meningioma task. In task 1, we achieved a DSC of 0.7900, 0.8076, 0.7760, 0.8926, 0.7874, 0.8938 and a HD95 of 35.63, 30.35, 44.58, 16.87, 38.19, 17.95 for ET, NETC, RC, SNFH, TC and WT, respectively and, in task 3, we achieved a DSC of 0.801 and HD95 of 38.26, in the testing phase. The code for these tasks is available at https://github.com/ShadowTwin41/BraTS_2023_2024_solutions.

Paper Structure

This paper contains 13 sections, 4 figures, 4 tables.

Table of Contents

  1. Introduction
  2. State of the art
  3. Methods
  4. Datasets
  5. Task 1 - Adult Glioma Post Treatment:
  6. Task 3 - Meningioma Radiotherapy:
  7. Data augmentation
  8. Networks
  9. Post-processing and Ensemble strategy
  10. Results
  11. Discussion and Conclusion
  12. Acknowledgments.
  13. Disclosure of Interests.

Figures (4)

  • Figure 1: The training pipeline of the GliGAN. The noise scan (z) and the label (y) are concatenated and fed into the generator (G). The discriminator (D) assesses the realism of a real scan and the reconstruction. Figure taken from ferreira2024we.
  • Figure 2: First row: Sample 00005-100 of the training set of Task 1 with a synthetic tumour in the healthy part of the brain in all 4 modalities and the corresponding segmentation. Second row: sample 0002-1 of the training set of Task 3 with a synthetic tumour in the healthy part of the brain and the corresponding segmentation.
  • Figure 3: Architectural design of the MedNeXt roy2023mednext.
  • Figure 4: Swin UNETR's architecture overview hatamizadeh2021swin.