Unsupervised learning of acquisition variability in structural connectomes via hybrid latent space modeling

2026-05-13

AI Summary

This study addresses the significant confounding effects of site, scanner, and acquisition protocol differences in multi-site diffusion MRI (dMRI) structural connectomes, which often obscure true biological variability. To tackle this challenge, the authors propose an unsupervised variational autoencoder framework that adaptively disentangles acquisition-related artifacts from biological signals by modeling a hybrid continuous-discrete latent space. A novel architecture-level annealing mechanism is introduced to dynamically balance the contributions of the two latent variable types at the encoder output, eliminating the need for manual hyperparameter tuning. Evaluated on a large multi-site dataset comprising 7,416 subjects, the method significantly outperforms baseline approaches (ARI = 0.53, p < 0.05) and effectively recovers clustering structures aligned with scanner and protocol configurations.

Abstract

Acquisition differences across sites, scanners, and protocols in dMRI introduce variability that complicates structural connectome analysis. This motivates deep learning models that can represent high-dimensional connectomes in a low-dimensional space while explicitly separating acquisition-related effects from biological variation. Conventional dimensionality reduction methods model all variance as continuous, so acquisition effects often get absorbed into a continuous latent space. Recent hybrid latent-space models combine discrete and continuous components to address this, but typically require manual capacity tuning to ensure the discrete component captures the intended variability. We introduce an unsupervised framework that removes this manual tuning by architecturally annealing encoder outputs before decoding, allowing the model to adaptively balance discrete and continuous latent variables during training. To evaluate it, we curated a dataset of N=7,416 structural connectomes derived from dMRI, spanning ages 2 to 102 and 13 studies with 25 unique acquisition-parameter combinations. Of these, 5,900 are cognitively unimpaired, 877 have mild cognitive impairment (MCI), and 639 have Alzheimer's disease (AD). We compare against a standard VAE, PCA with k-means clustering, and hybrid models that anneal only through the loss function. Our architectural annealing produces stronger site learning (ARI=0.53, p<0.05) than these baselines. Results show that a hybrid continuous-discrete latent space, with architectural rather than loss-based annealing, provides a useful unsupervised mechanism for capturing acquisition variability in dMRI: by jointly modeling smooth and categorical structure, the Joint-VAE recovers clusters aligned with scanner and protocol differences.

Problem

• structural connectomes
• acquisition variability
• dMRI
• unsupervised learning
• latent space

Innovation

• hybrid latent space
• architectural annealing
• unsupervised disentanglement
• structural connectome
• dMRI acquisition variability