Explaining anatomical shape variability: supervised disentangling with a variational graph autoencoder
Faculty Advisor
Date
2023
Keywords
anatomical shape analysis, graph convolution, hippocampus, latent space disentanglement
Abstract (summary)
This work proposes a modular geometric deep learning framework that isolates shape variability associated with a given scalar factor (e.g., age) within a population (e.g., healthy individuals). Our approach leverages a novel graph convolution operator in a variational autoencoder to process 3D mesh data and learn a meaningful, low-dimensional shape descriptor. A supervised disentanglement strategy aligns a single component of this descriptor with the factor of interest during training. On a toy synthetic dataset and a high-resolution diffusion tensor imaging (DTI) dataset, the proposed model is better able to disentangle the learned latent space with a simulated factor and patient age, respectively, relative to other state-of-the-art methods. The relationship between age and shape estimated in the DTI analysis is consistent with existing neuroimaging literature.
Publication Information
DOI
Notes
Presented on April 18, 2023, at the IEEE 20th International Symposium on Biomedical Imaging (ISBI) Conference in Cartagena, Colombia.
Item Type
Article
Language
Rights
All Rights Reserved