Decoder-Free Supervoxel GNN for Accurate Brain-Tumor Localization in Multi-Modal MRI

TL;DR

This paper introduces SVGFormer, a decoder-free, graph-based neural network for 3D brain tumor localization in multi-modal MRI, emphasizing feature learning and interpretability over traditional encoder-decoder architectures.

Contribution

Proposes SVGFormer, a novel decoder-free supervoxel graph neural network that enhances feature encoding and explainability in 3D medical imaging tasks.

Findings

Achieved F1-score of 0.875 in tumor classification

Attained MAE of 0.028 in tumor proportion regression

Demonstrated strong performance and interpretability of the model

Abstract

Modern vision backbones for 3D medical imaging typically process dense voxel grids through parameter-heavy encoder-decoder structures, a design that allocates a significant portion of its parameters to spatial reconstruction rather than feature learning. Our approach introduces SVGFormer, a decoder-free pipeline built upon a content-aware grouping stage that partitions the volume into a semantic graph of supervoxels. Its hierarchical encoder learns rich node representations by combining a patch-level Transformer with a supervoxel-level Graph Attention Network, jointly modeling fine-grained intra-region features and broader inter-regional dependencies. This design concentrates all learnable capacity on feature encoding and provides inherent, dual-scale explainability from the patch to the region level. To validate the framework's flexibility, we trained two specialized models on the BraTS dataset: one for node-level classification and one for tumor proportion regression. Both models achieved strong performance, with the classification model achieving a F1-score of 0.875 and the regression model a MAE of 0.028, confirming the encoder's ability to learn discriminative and localized features. Our results establish that a graph-based, encoder-only paradigm offers an accurate and inherently interpretable alternative for 3D medical image representation.