From Baseline to Follow-Up: Counterfactual Spine DXA Image Synthesis in UK Biobank Using a Causal Hierarchical Variational Autoencoder

TL;DR

This paper introduces a causal hierarchical variational autoencoder for generating anatomically consistent DXA spine images from UK Biobank data, enabling counterfactual analysis of anatomical changes over time.

Contribution

It presents a novel causal generative model conditioned on metadata for anatomically plausible DXA image synthesis and longitudinal analysis.

Findings

Strong agreement in vertebral morphometry variables after age intervention

Model achieves causally consistent counterfactual image generation

Supports intervention-aligned synthesis of DXA images

Abstract

Dual-energy X-ray absorptiometry (DXA) is widely used for large-scale skeletal assessment, yet learning controllable and interpretable factor-specific anatomical variation remains challenging. We propose a metadata-conditioned causal hierarchical variational autoencoder (CHVAE) for causally consistent generation of anteroposterior (AP) spine DXA images from the UK Biobank (UKB). The model is trained on 3,743 raw AP spine scans from the first imaging visit and conditioned on basic participant attributes and lumbar morphometry. Causal consistency is evaluated in a baseline-to-follow-up setting using abduction--action--prediction (AAP): latent variables are abducted from baseline images, age is intervened to the repeat-imaging value, and the resulting counterfactual follow-up morphometry is compared with observed repeat-imaging measurements. Results show strong absolute-level agreement for key vertebral morphometry variables under age intervention, supporting intervention-aligned synthesis of anatomically plausible DXA images.