Metadata Supervised MRI Representations for Modelling and Controlling Acquisition Variability

2026-07-13

AI Summary

In magnetic resonance imaging, variability introduced by differences in scanners and acquisition protocols often confounds models by entangling anatomical structure with imaging appearance, thereby compromising generalization and interpretability. This work proposes a decoupled representation learning approach supervised by DICOM metadata, which— for the first time—explicitly models acquisition-induced variability as an auditable and controllable structured component. By jointly modeling images and their associated metadata through deep learning, the method effectively disentangles anatomical content from contrast-related appearance. The resulting unified image harmonization model efficiently organizes heterogeneous imaging sequences, facilitates sequence interpretation, detects image–metadata inconsistencies, and suppresses acquisition-specific variations while preserving biologically meaningful information.

Abstract

Magnetic resonance imaging exhibits substantial acquisition variability, where identical anatomy can appear markedly different across scanners and imaging protocols. Consequently, learned representations entangle biological structure with acquisition-dependent appearance, limiting interpretability, generalisation, and clinical deployment. We show that these sources of variation can be separated by jointly modelling MRI images and DICOM metadata. Using large-scale clinical brain MRI data, we learn representations that separate anatomical structure from contrast-dependent appearance. Resulting contrast representations organise heterogeneous acquisitions, support sequence understanding, and detect image--metadata inconsistencies, whereas anatomical representations suppress acquisition-specific variation while preserving biologically relevant information. Building on these disentangled representations, we introduce a unified anatomy-preserving harmonisation model for cross-modality and cross-site adaptation, conditioned on image or acquisition metadata. Our findings suggest that acquisition variability is a structured component of the imaging process that can be modelled, audited, and controlled, providing a foundation for acquisition-aware representation learning in large-scale medical imaging.

Problem

• acquisition variability
• MRI representations
• DICOM metadata
• image harmonisation
• contrast-dependent appearance

Innovation

• disentangled representation
• metadata-supervised learning
• MRI harmonisation
• acquisition variability
• DICOM metadata