Disentangling Hippocampal Shape Variations: A Study of Neurological Disorders Using Mesh Variational Autoencoder with Contrastive Learning

Jakaria Rabbi1Orcid, Johannes Kiechle2Orcid, Christian Beaulieu3Orcid, Nilanjan Ray1Orcid, Dana Cobzas4Orcid
1: Department of Computing Science, University of Alberta, Edmonton, Canada, 2: Institute for Computational Imaging and AI in Medicine, Technical University of Munich, Munich, Germany, 3: Department of Radiology and Diagnostic Imaging & Biomedical Engineering, University of Alberta, Edmonton, Canada, 4: Department of Computer Science, MacEwan University, Edmonton, Canada
Publication date: 2024/11/25
https://doi.org/10.59275/j.melba.2024-267f
PDF · Code · arXiv

Abstract

This paper presents a comprehensive study focused on disentangling hippocampal shape variations from diffusion tensor imaging (DTI) datasets within the context of neurological disorders. Leveraging a Mesh Variational Autoencoder (VAE) enhanced with Supervised Contrastive Learning, our approach aims to improve interpretability by disentangling two distinct latent variables corresponding to age and the presence of diseases. In our ablation study, we investigate a range of VAE architectures and contrastive loss functions, showcasing the enhanced disentanglement capabilities of our approach. This evaluation uses synthetic 3D torus mesh data and real 3D hippocampal mesh datasets derived from the DTI hippocampal dataset. Our supervised disentanglement model outperforms several state-of-the-art (SOTA) methods like attribute and guided VAEs in terms of disentanglement scores. Our model distinguishes between age groups and disease status in patients with Multiple Sclerosis (MS) using the hippocampus data. Our Mesh VAE with Supervised Contrastive Learning shows the volume changes of the hippocampus of MS populations at different ages, and the result is consistent with the current neuroimaging literature. This research provides valuable insights into the relationship between neurological disorder and hippocampal shape changes in different age groups of MS populations using a Mesh VAE with Supervised Contrastive loss.
Our code is available at https://github.com/Jakaria08/Explaining_Shape_Variability

Keywords

Disentangled Representation Learning · Mesh Variational Autoencoder · Deep Learning · Contrastive Learning · Neurological Disorders · Medical Imaging · Hippocampal Shape Variations · Diffusion Tensor Image

Bibtex @article{melba:2024:030:rabbi, title = "Disentangling Hippocampal Shape Variations: A Study of Neurological Disorders Using Mesh Variational Autoencoder with Contrastive Learning", author = "Rabbi, Jakaria and Kiechle, Johannes and Beaulieu, Christian and Ray, Nilanjan and Cobzas, Dana", journal = "Machine Learning for Biomedical Imaging", volume = "2", issue = "November 2024 issue", year = "2024", pages = "2268--2292", issn = "2766-905X", doi = "https://doi.org/10.59275/j.melba.2024-267f", url = "https://melba-journal.org/2024:030" }
RISTY - JOUR AU - Rabbi, Jakaria AU - Kiechle, Johannes AU - Beaulieu, Christian AU - Ray, Nilanjan AU - Cobzas, Dana PY - 2024 TI - Disentangling Hippocampal Shape Variations: A Study of Neurological Disorders Using Mesh Variational Autoencoder with Contrastive Learning T2 - Machine Learning for Biomedical Imaging VL - 2 IS - November 2024 issue SP - 2268 EP - 2292 SN - 2766-905X DO - https://doi.org/10.59275/j.melba.2024-267f UR - https://melba-journal.org/2024:030 ER -

2024:030 cover