Surface Vision Transformers: Flexible Attention-Based Modelling of Biomedical Surfaces
This work addresses the challenge of modeling biomedical surfaces for tasks like brain age prediction and disease classification, offering a novel approach that could benefit medical imaging analysis, though it is incremental in extending ViTs to a new domain.
The authors tackled the problem of applying Vision Transformers to biomedical surface data by reformulating surface learning as a sequence-to-sequence task, achieving consistent improvements over geometric deep learning methods for brain age and fluid intelligence prediction and comparable performance on calcium score classification to clinical standards.
Recent state-of-the-art performances of Vision Transformers (ViT) in computer vision tasks demonstrate that a general-purpose architecture, which implements long-range self-attention, could replace the local feature learning operations of convolutional neural networks. In this paper, we extend ViTs to surfaces by reformulating the task of surface learning as a sequence-to-sequence learning problem, by proposing patching mechanisms for general surface meshes. Sequences of patches are then processed by a transformer encoder and used for classification or regression. We validate our method on a range of different biomedical surface domains and tasks: brain age prediction in the developing Human Connectome Project (dHCP), fluid intelligence prediction in the Human Connectome Project (HCP), and coronary artery calcium score classification using surfaces from the Scottish Computed Tomography of the Heart (SCOT-HEART) dataset, and investigate the impact of pretraining and data augmentation on model performance. Results suggest that Surface Vision Transformers (SiT) demonstrate consistent improvement over geometric deep learning methods for brain age and fluid intelligence prediction and achieve comparable performance on calcium score classification to standard metrics used in clinical practice. Furthermore, analysis of transformer attention maps offers clear and individualised predictions of the features driving each task. Code is available on Github: https://github.com/metrics-lab/surface-vision-transformers