Introduction
Stroke is a leading cause of long-term adult disability \cite{Feigin_2014}. Despite the significant public health burden of stroke \cite{Di_Carlo_2008}, the ability to individually prognosticate stroke outcomes in the acute setting is difficult. Prediction of long-term functional outcomes in patients with acute ischemic stroke (AIS) is important as it may offer the possibility of personalized early interventions and rehabilitation strategies. However, this prediction remains challenging \cite{Counsell_2001} due to the complex mechanisms of post-stroke recovery and the multitude of clinical and radiographic variables that differentially affect patient outcomes \cite{Kent_2001,Meschia_2002,Muir_2002}. Associations between structural features, such as white matter microstructural integrity, and functional post-stroke outcome have recently been established \cite{Etherton_2017}. However, the effect of structural or functional brain connectivity organisation on recovery after stroke and its role in resilience to damage is yet to be fully elucidated.
Recent methodological developments in medical imaging enabled high level representations of a subject's underlying biology. In particular, connectomics involves representation of the brain as a graph and allows the exploration of topological properties of brain connectivity with network theoretical measures \cite{Rubinov_2010}. Graph theory provides many valuable tools for the study of the human brain and can lead to fundamental insights into its organisation \cite{van_den_Heuvel_2011,Chung_2016,daniel2016} and how it may be altered due to disease \cite{Fornito_2015,Kawahara_2017,Ktena_2018}. At the same time, it incorporates knowledge of elementary system components (i.e. network nodes or brain regions) as well as the interactions between them (i.e. network edges or functional and/or structural connections) and analyzes the properties of the emerging complex system. This network-centric perspective can provide insights into how the brain's resilient network architecture allows it to withstand injury \cite{Achard_2006,Joyce_2013,Lo_2015}. Moreover, network science can elucidate the mechanisms through which diseases affect brain regions and progress over time \cite{Raj_2012,Deco_2015}.
A so-called rich club organization has been described \cite{van_den_Heuvel_2011} in the human connectome, which comprises a set of network hubs that tend to be more densely connected than expected by chance, given their high network degree, and has been explored in healthy individuals as well as patients with psychiatric disorders \cite{van_den_Heuvel_2013}. These hubs are thought to serve as a high capacity backbone critical for physiological neuronal connectivity whose integrity is crucial for brain function. Targeted attacks on their connections can, thus, have a significant impact on global network efficiency \cite{van_den_Heuvel_2011}. At the same time, stroke location has been identified as an independent predictor of cognitive outcome \cite{Wu_2015,Munsch_2015}, in addition to widely accepted clinical predictors, such as age \cite{Jokinen_2015} and stroke lesion size \cite{Vogt_2012a,Hope_2013}.