Brain connectivity geometry controls classification and transfer accuracy
It's the Geometry, Not the Model: Effective Rank and Subspace Alignment in Functional Connectivity Classification
Computational GeometryComputer Vision and Pattern Recognition
Summary
Classifying brain states using resting brain activity often assumes using all connections helps. This paper shows that the shape, or geometry, of brain connectivity data matters more than the specific model used. The authors found that most variation lies in a small space of key patterns, but these patterns differ across research sites, making it hard to transfer classification results. They suggest that matching these key patterns across sites is crucial for better brain diagnosis using this method.
What this means in practice
- •For hospital data teams: Improve brain disorder diagnosis tools by aligning connectivity data subspaces from different hospitals to enhance classification accuracy across sites.
- •For medical imaging software developers: Create preprocessing modules that adjust brain connectivity data geometry to enable consistent classification results when combining data from multiple imaging centers.
Authors
Xiao Fan, Jingyuan Li, Yubo Han, Hongbin Guo, Guanya Li, Yang Hu, Wenchao Zhang, Weibin Ji, Yi Zhang
Abstract
Resting-state functional connectivity (FC) is widely used to classify brain phenotypes and disorders. Most pipelines use the full connectome and seek gains through model design. We instead examine how FC geometry constrains classification and cross-site transfer. Across-subject FC variation concentrates in a small effective subspace, suggesting substantial redundancy in nominal dimensions. Across cohorts, these subspaces may differ in orientation even when their effective ranks are comparable, potentially limiting transfer. Across 2,330 subjects from HCP, ABIDE, and ADHD-200, effective-rank analysis reveals strong spectral concentration. Projection onto leading components at the effective-rank scale recovers most of the full-FC classification performance. In ABIDE, site-specific effective subspaces are weakly aligned, and their principal-angle overlap predicts pairwise transfer after covariate adjustment despite comparable per-site effective ranks. Controlled rotations that alter subspace orientation while preserving the mean and covariance spectrum drive transfer toward chance, whereas displacement-matched label-orthogonal rotations do not. These results identify subspace orientation as a key factor in transfer degradation under controlled perturbations. This study offers a geometric diagnostic of FC generalization and suggests evaluating cross-site harmonization by its ability to align effective subspaces alongside classification accuracy.