Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Generative Latent Representations of 3D Brain MRI for Multi-Task Downstream Analysis in Down Syndrome

Jordi Malé, Juan Fortea, Mateus Rozalem-Aranha, Neus Martínez-Abadías, Xavier Sevillano

TL;DR

This work develops multiple variational autoencoders (VAEs) to encode 3D brain MRI scans into compact latent space representations for generative and predictive applications and demonstrates that the VAE successfully captures essential brain features while maintaining high reconstruction fidelity.

Abstract

Generative models have emerged as powerful tools in medical imaging, enabling tasks such as segmentation, anomaly detection, and high-quality synthetic data generation. These models typically rely on learning meaningful latent representations, which are particularly valuable given the high-dimensional nature of 3D medical images like brain magnetic resonance imaging (MRI) scans. Despite their potential, latent representations remain underexplored in terms of their structure, information content, and applicability to downstream clinical tasks. Investigating these representations is crucial for advancing the use of generative models in neuroimaging research and clinical decision-making. In this work, we develop multiple variational autoencoders (VAEs) to encode 3D brain MRI scans into compact latent space representations for generative and predictive applications. We systematically evaluate the effectiveness of the learned representations through three key analyses: (i) a quantitative and qualitative assessment of MRI reconstruction quality, (ii) a visualisation of the latent space structure using Principal Component Analysis, and (iii) downstream classification tasks on a proprietary dataset of euploid and Down syndrome individuals brain MRI scans. Our results demonstrate that the VAE successfully captures essential brain features while maintaining high reconstruction fidelity. The latent space exhibits clear clustering patterns, particularly in distinguishing individuals with Down syndrome from euploid controls.

Generative Latent Representations of 3D Brain MRI for Multi-Task Downstream Analysis in Down Syndrome

TL;DR

This work develops multiple variational autoencoders (VAEs) to encode 3D brain MRI scans into compact latent space representations for generative and predictive applications and demonstrates that the VAE successfully captures essential brain features while maintaining high reconstruction fidelity.

Abstract

Generative models have emerged as powerful tools in medical imaging, enabling tasks such as segmentation, anomaly detection, and high-quality synthetic data generation. These models typically rely on learning meaningful latent representations, which are particularly valuable given the high-dimensional nature of 3D medical images like brain magnetic resonance imaging (MRI) scans. Despite their potential, latent representations remain underexplored in terms of their structure, information content, and applicability to downstream clinical tasks. Investigating these representations is crucial for advancing the use of generative models in neuroimaging research and clinical decision-making. In this work, we develop multiple variational autoencoders (VAEs) to encode 3D brain MRI scans into compact latent space representations for generative and predictive applications. We systematically evaluate the effectiveness of the learned representations through three key analyses: (i) a quantitative and qualitative assessment of MRI reconstruction quality, (ii) a visualisation of the latent space structure using Principal Component Analysis, and (iii) downstream classification tasks on a proprietary dataset of euploid and Down syndrome individuals brain MRI scans. Our results demonstrate that the VAE successfully captures essential brain features while maintaining high reconstruction fidelity. The latent space exhibits clear clustering patterns, particularly in distinguishing individuals with Down syndrome from euploid controls.
Paper Structure (18 sections, 1 equation, 4 figures, 7 tables)

This paper contains 18 sections, 1 equation, 4 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Diagnostic Classification based on Latent Space Representations
  3. Learning Latent Space Representations via Variational Autoencoders
  4. Latent Space-based Diagnostic Classification
  5. Experimental Setup
  6. Datasets Description and Preprocessing.
  7. Classification Tasks.
  8. Implementation Details.
  9. Results and Discussion
  10. VAE Reconstruction Quality.
  11. VAE Reconstruction Fidelity.
  12. Latent Space Visualization via PCA.
  13. EU vs DS classification.
  14. AD Stage Classification in DS Subjects.
  15. ID Classification in DS Subjects.
  16. ...and 3 more sections

Figures (4)

  • Figure 1: Overview of the proposed pipeline for latent space generation and disease classification. The framework consists of two main components: A) a VAE that learns latent representations from 3D brain MRI scans, and B) a classification module that utilises these latent representations to predict a diagnosis or condition.
  • Figure 2: Preprocessing pipeline for MRI scans, including bias correction, two-level affine registration and skull-stripping.
  • Figure 3: Reconstruction of a 3D brain MRI using the trained VAE (brain reconstruction selected randomly).
  • Figure 4: PCA visualisation of the latent representations learned by the VAE trained on euploid subjects from the IXI dataset and the HCP dataset.