Shape-Based Inductive Bias for Glioma Grading from Tumor Contours

2026-07-27

AI Summary

This study addresses the limitations in glioma grading caused by neglecting tumor contour shape and overreliance on pixel-level modeling. To overcome these issues, the authors propose a lightweight, function-based shape alignment method that decomposes tumor contours into global deformations and residual Fourier shape components. The resulting contours are encoded as frequency-ordered tokens and classified using a compact multilayer perceptron (MLP). By incorporating shape-aware inductive bias at the representation level, this approach drastically reduces feature dimensionality and model parameters—by at least 46× compared to ResNet-18 and ViT-Tiny—while enhancing interpretability and scalability. Evaluated on the BraTS 2020 dataset, the method achieves an average balanced accuracy of 71.5% and an F1 score of 54.9% for low-grade gliomas.

Abstract

Glioma grading from tumor contours is often treated as a pixel problem even when the signal of interest is shape. We align closed contours with a functional shape-alignment framework, separate global deformation from residual Fourier shape, and organize these quantities as frequency-ordered tokens. In five-fold patient-disjoint cross-validation on BraTS~2020 tumor contours, with model selection performed using grouped inner validation, a compact multilayer perceptron (MLP) achieves the highest mean balanced accuracy at 71.5\%, compared with 65.9\% for ResNet-18 and 63.3\% for ViT-Tiny. It also gives the highest mean low-grade glioma F1 at 54.9\%. Its pooled out-of-fold balanced accuracy is 72.4\% (patient-bootstrap 95\% CI: 66.4--77.8\%). The selected MLPs use 2.9k--117.3k parameters across folds, at least 46 times fewer than the pixel baselines. In a controlled noise-free simulation, shape-based models reach 56.3--71.5\% balanced accuracy while the pixel models remain at 50.0--52.5\%. This work demonstrates how incorporating a shape-based inductive bias at the representation level can improve interpretability and scalability while enabling substantial dimensionality reduction.

Problem

• glioma grading
• tumor contours
• shape-based inductive bias
• shape representation
• medical image analysis

Innovation

• shape-based inductive bias
• functional shape alignment
• Fourier shape representation
• frequency-ordered tokens
• compact MLP