Improving Normative Modeling for Multi-modal Neuroimaging Data using mixture-of-product-of-experts variational autoencoders

TL;DR

This paper introduces a novel variational autoencoder approach using mixture-of-product-of-experts to improve normative modeling of multimodal neuroimaging data, enabling better detection of disease deviations and associated brain regions.

Contribution

The study proposes a Mixture-of-Product-of-Experts VAE that enhances joint latent space modeling for normative neuroimaging, improving disease deviation detection and interpretability.

Findings

Better modeling of joint latent distributions in multimodal data

Improved detection of outliers related to Alzheimer's Disease

Identification of brain regions linked to abnormal deviations

Abstract

Normative models in neuroimaging learn the brain patterns of healthy population distribution and estimate how disease subjects like Alzheimer's Disease (AD) deviate from the norm. Existing variational autoencoder (VAE)-based normative models using multimodal neuroimaging data aggregate information from multiple modalities by estimating product or averaging of unimodal latent posteriors. This can often lead to uninformative joint latent distributions which affects the estimation of subject-level deviations. In this work, we addressed the prior limitations by adopting the Mixture-of-Product-of-Experts (MoPoE) technique which allows better modelling of the joint latent posterior. Our model labelled subjects as outliers by calculating deviations from the multimodal latent space. Further, we identified which latent dimensions and brain regions were associated with abnormal deviations due to AD pathology.