A Convolutional Autoencoder Approach to Learn Volumetric Shape Representations for Brain Structures

TL;DR

This paper introduces a convolutional autoencoder that learns invariant volumetric shape representations of brain structures directly from MRI segmentation images, improving shape retrieval accuracy without surface preprocessing.

Contribution

The novel autoencoder-based method automatically learns robust, transformation-invariant shape descriptors from volumetric data, outperforming existing benchmarks in neuroanatomical shape retrieval.

Findings

Outperforms state-of-the-art shape retrieval benchmarks

Automatically enhances inter-subject shape differences

Invariance to affine transformations achieved

Abstract

We propose a novel machine learning strategy for studying neuroanatomical shape variation. Our model works with volumetric binary segmentation images, and requires no pre-processing such as the extraction of surface points or a mesh. The learned shape descriptor is invariant to affine transformations, including shifts, rotations and scaling. Thanks to the adopted autoencoder framework, inter-subject differences are automatically enhanced in the learned representation, while intra-subject variances are minimized. Our experimental results on a shape retrieval task showed that the proposed representation outperforms a state-of-the-art benchmark for brain structures extracted from MRI scans.