Related Experiment Video
Updated: Jul 21, 2025
Automated Segmentation of Cortical Grey Matter from T1-Weighted MRI Images
Published on: January 7, 2019
Meningioma Segmentation in T1-Weighted MRI Leveraging Global Context and Attention Mechanisms
David Bouget1, André Pedersen1, Sayied Abdol Mohieb Hosainey2
1Department of Health Research, SINTEF Digital, Trondheim, Norway.
Abstract:
Purpose: Meningiomas are the most common type of primary brain tumor, accounting for ~30% of all brain tumors. A substantial number of these tumors are never surgically removed but rather monitored over time. Automatic and precise meningioma segmentation is, therefore, beneficial to enable reliable growth estimation and patient-specific treatment planning. Methods: In this study, we propose the inclusion of attention mechanisms on top of a U-Net architecture used as backbone: (i) Attention-gated U-Net (AGUNet) and (ii) Dual Attention U-Net (DAUNet), using a three-dimensional (3D) magnetic resonance imaging (MRI) volume as input. Attention has the potential to leverage the global context and identify features' relationships across the entire volume. To limit spatial resolution degradation and loss of detail inherent to encoder-decoder architectures, we studied the impact of multi-scale input and deep supervision components. The proposed architectures are trainable end-to-end and each concept can be seamlessly disabled for ablation studies. Results: The validation studies were performed using a five-fold cross-validation over 600 T1-weighted MRI volumes from St. Olavs Hospital, Trondheim University Hospital, Norway. Models were evaluated based on segmentation, detection, and speed performances, and results are reported patient-wise after averaging across all folds. For the best-performing architecture, an average Dice score of 81.6% was reached for an F1-score of 95.6%. With an almost perfect precision of 98%, meningiomas smaller than 3 ml were occasionally missed hence reaching an overall recall of 93%. Conclusion: Leveraging global context from a 3D MRI volume provided the best performances, even if the native volume resolution could not be processed directly due to current GPU memory limitations. Overall, near-perfect detection was achieved for meningiomas larger than 3 ml, which is relevant for clinical use. In the future, the use of multi-scale designs and refinement networks should be further investigated. A larger number of cases with meningiomas below 3 ml might also be needed to improve the performance for the smallest tumors.
Insights
This study introduces attention-based U-Net models for precise meningioma segmentation in 3D MRI scans. The best model achieved 81.6% Dice score, enabling reliable tumor monitoring and treatment planning.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Neuro-oncology
Background:
- Meningiomas are the most common primary brain tumors, often monitored non-surgically.
- Accurate segmentation is crucial for tracking tumor growth and planning patient-specific treatments.
Purpose of the Study:
- To develop and evaluate attention-based U-Net architectures (AGUNet, DAUNet) for automatic and precise meningioma segmentation in 3D MRI.
- To investigate the impact of multi-scale input and deep supervision on segmentation accuracy and detail preservation.
Main Methods:
- Proposed attention-gated U-Net (AGUNet) and dual attention U-Net (DAUNet) architectures using 3D MRI volumes.
- Incorporated multi-scale input and deep supervision to mitigate resolution loss.
- Trained and validated models end-to-end on 600 T1-weighted MRI scans using five-fold cross-validation.
Main Results:
- The best-performing model achieved an average Dice score of 81.6% and an F1-score of 95.6%.
- Achieved high precision (98%) and recall (93%), with near-perfect detection for meningiomas larger than 3 ml.
- Attention mechanisms improved segmentation by leveraging global context within the 3D MRI volume.
Conclusions:
- Attention-based deep learning models significantly enhance meningioma segmentation accuracy in 3D MRI.
- The developed models show clinical relevance for monitoring larger meningiomas, with potential for improvement in segmenting smaller tumors.
- Future work should explore multi-scale designs and refinement networks for enhanced performance, particularly for small tumors.
