English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Open

The development of robust and generalisable models for encoding thespatio-temporal dynamics of human brain activity is crucial for advancingneuroscientific discoveries. However, significant individual variation in theorganisation of the human cerebral cortex makes it difficult to identifypopulation-level trends in these signals. Recently, Surface Vision Transformers(SiTs) have emerged as a promising approach for modelling cortical signals, yetthey face some limitations in low-data scenarios due to the lack of inductivebiases in their architecture. To address these challenges, this paper proposesthe surface Masked AutoEncoder (sMAE) and video surface Masked AutoEncoder(vsMAE) - for multivariate and spatio-temporal pre-training of cortical signalsover regular icosahedral grids. These models are trained to reconstructcortical feature maps from masked versions of the input by learning stronglatent representations of cortical structure and function. Such representationstranslate into better modelling of individual phenotypes and enhancedperformance in downstream tasks. The proposed approach was evaluated oncortical phenotype regression using data from the young adult Human ConnectomeProject (HCP) and developing HCP (dHCP). Results show that (v)sMAE pre-trainedmodels improve phenotyping prediction performance on multiple tasks by $\ge26\%$, and offer faster convergence relative to models trained from scratch.Finally, we show that pre-training vision transformers on large datasets, suchas the UK Biobank (UKB), supports transfer learning to low-data regimes. Ourcode and pre-trained models are publicly available athttps://github.com/metrics-lab/surface-masked-autoencoders .

Spatio-Temporal Encoding of Brain Dynamics with Surface Masked Autoencoders