A Convolutional Autoencoder Approach to Learn Volumetric Shape Representations for Brain Structures
This work addresses shape analysis for brain structures in neuroimaging, offering an incremental improvement over existing methods.
The authors tackled the problem of studying neuroanatomical shape variation by proposing a convolutional autoencoder that learns volumetric shape representations from binary segmentation images without pre-processing, achieving state-of-the-art performance on a shape retrieval task for brain structures from MRI scans.
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.