The relationship between brain damage and functional outcomes is difficult but important to predict. After ischaemic stroke, there is wide variability immediate impairment and prospective recovery. I began my work at KCL with Dr Mike O'Sullivan on a comparatively large, clinical, observational study, looking at cognitive status over the first year after stroke. We related the status of specific white matter structures to memory impairment using state of the art MRI techniques and guided by specific hypotheses about recovery of function based on white matter reorganisation. My current work looks at clinical brain imaging data, which is less specialised but available in vastly greater quantities, on the hypothesis that low-precision, high-dimensional data can be have greater predictive power than high-precision, low-dimensional data. I work alongside software engineers and machine learning specialists in the AMIGO team led by Jorge Cardoso in the School of Biomedical Engineering and Imaging Sciences.

Neuroimaging biomarkers for cognitive recovery after stroke.

• fMRI
• DTI
• stroke
• episodic memory
• neuroanatomy
• multivariate analysis
• machine learning

My PhD research at the University of Florida developed fMRI paradigms to investigate fine-grained aspects of emotion processing. These paradigms were later applied to studiy patients with depression.

In 2007 I returned to the UK to work as a post-doc at the Centre for Speech, Language and the Brain at the University of Cambridge (http://csl.psychol.cam.ac.uk). As part of a large MRC-funded project investigating the neural basis for spoken language comprehension I analysed cognitive and neuroimaging data from healthy individuals and stroke survivors. My particular interest was in applying novel imaging analyses to tease apart the relationships between structure, function and different aspects of spoken language comprehension.

My second project at the CSLB investigated perceptual and conceptual processing and included individuals with damage to the ventral temporal cortex. To test the CSLB's neurocognitive model of conceptual processing, I related the local anatomy of key anterior temporal regions with their structural connections and implmented analyses relating structuring connectivity to local functional activity.

I joined KCL in 2014 to further my interest in relating neuroimaging measures to outcomes in patients with brain damage. I worked first on the STRATEGIC study with Dr Mike O'Sullivan, relating neuroimaging to cognitive measures over the first year after stroke. I now work with the AMIGO team, led by Dr Jorge Cardoso, developing machine learning tools that use large clinical datasets to predict outcomes and improve diagnosis.