Abstract Resting-state functional magnetic resonance imaging ( rs-fmri ) is widely used in connectomics for studying the functional relationships between regions of the human brain. rs-fmri connectomics, however, has inherent analytical challenges, such as accounting for negative correlations. In addition, functional relationships between brain regions do not necessarily correspond to their anatomical distance, making the intrinsic geometry of the functional connectome less well understood. Recent techniques in natural language processing and machine learning, such as word2vec, have used embedding methods to map high-dimensional data into meaningful vector spaces. Inspired by this approach, we have developed a graph embedding pipeline, rest2vec, for studying the intrinsic geometry of functional connectomes. We demonstrate how rest2vec uses the phase angle spatial embedding ( phase ) method with dimensionality reduction techniques to embed the functional connectome into lower dimensions. Rest2vec can also be linked to the maximum mean discrepancy ( mmd ) metric to assign functional modules of the connectome in a continuous manner, improving upon traditional binary classification methods. Together, this allows for studying the functional connectome such that the full range of correlative information is preserved and gives a more informed understanding of the functional organization of the brain.

rest2vec: Vectorizing the resting-state functional connectome using graph embedding