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

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Protani, Andrea, Bosch, Marc Molina Van Den, Giusti, Lorenzo, Da Silva, Heloisa Barbosa, Cacace, Paolo, Aillet, Albert Sund, Ballester, Miguel Angel Gonzalez, Hummel, Friedhelm, Serio, Luigi
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911387254521856
author Protani, Andrea
Bosch, Marc Molina Van Den
Giusti, Lorenzo
Da Silva, Heloisa Barbosa
Cacace, Paolo
Aillet, Albert Sund
Ballester, Miguel Angel Gonzalez
Hummel, Friedhelm
Serio, Luigi
author_facet Protani, Andrea
Bosch, Marc Molina Van Den
Giusti, Lorenzo
Da Silva, Heloisa Barbosa
Cacace, Paolo
Aillet, Albert Sund
Ballester, Miguel Angel Gonzalez
Hummel, Friedhelm
Serio, Luigi
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2601_14055
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Decoder-Free Supervoxel GNN for Accurate Brain-Tumor Localization in Multi-Modal MRI
Protani, Andrea
Bosch, Marc Molina Van Den
Giusti, Lorenzo
Da Silva, Heloisa Barbosa
Cacace, Paolo
Aillet, Albert Sund
Ballester, Miguel Angel Gonzalez
Hummel, Friedhelm
Serio, Luigi
Computer Vision and Pattern Recognition
Artificial Intelligence
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.
title Decoder-Free Supervoxel GNN for Accurate Brain-Tumor Localization in Multi-Modal MRI
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2601.14055