Lightweight vision transformer improves brain tumor segmentation accuracy
Lightweight Vision Transformer-Based U-Net for Brain Tumor Segmentation from MRI
Computer Vision and Pattern Recognition
Summary
Brain tumor segmentation from MRI scans is important for diagnosis and treatment but is challenging due to tumors' irregular shapes. The authors propose a new method that combines two techniques—UNet and Vision Transformers—to better capture both local details and the overall shape of tumors. Their model is efficient, with fewer parameters than usual, yet achieves better accuracy on a standard brain tumor dataset. This suggests the approach could help doctors by providing more precise tumor maps.
What this means in practice
- •For medical imaging teams: Segment brain tumors more accurately in MRI scans to improve diagnosis and treatment planning.
- •For clinical software developers: Integrate efficient transformer-based segmentation models into surgical guidance tools for brain tumor removal.$Commercial implications: Enables advanced segmentation features in surgical software marketed to hospitals for better tumor visualization.
Tested on one dataset.
Authors
Sheekar Banerjee, Md. Srabon Chowdhury, Md. Mahbub Hasan Akash, Ishtiak Al Mamoon
Abstract
Accurate brain tumor segmentation from Magnetic Resonance Imaging is essential for diagnosis, treatment planning, and surgical guidance. Although Convolutional Neural Networks, particularly UNet, have achieved significant success in medical image segmentation, they often struggle to capture the long-range spatial dependencies required to model tumors with irregular shapes and complex boundaries. This paper proposes a lightweight Vision Transformer UNet that combines the hierarchical feature extraction capability of UNet with the global context modeling of Vision Transformers. The proposed architecture incorporates a compact ViT bottleneck within a U-Net encoder-decoder framework, enabling effective learning of both local and global features while maintaining computational efficiency with only 2.6 million trainable parameters. The model was evaluated on the TCGA LGG MRI Segmentation dataset, achieving a mean Intersection over Union of 0.8100 and a Dice score of 0.8446, outperforming the baseline UNet by 3.75% and 3.15%, respectively. Extensive quantitative and qualitative analyses, including confusion matrix evaluation, precision recall curves, per-image performance distribution, and tumor size dependency analysis, demonstrate the effectiveness and robustness of the proposed method for brain tumor segmentation.