Sarah E. Morgan
Accelerate Science Research Fellow
Cambridge University
New network approaches to brain MRI and speech data
At the turn of the century, Prof Stephen Hawking wrote that the “21st century will be the century of complexity”. In mental health research, the ever increasing amounts of data available from genetics, genomics, brain imaging, electronic health records and wearables make it easy to understand why. Networks provide an invaluable tool to unravel data from different modalities, allowing us to bridge spatial and temporal scales and provide a more holistic view of mental health.
In this talk, I will share two examples of new network approaches to studying mental health data. In the first, we propose a new method to estimate structural similarity networks from brain MRI, termed ‘Morphometric INverse Divergence (MIND) networks’ (Sebenius et al, in preparation). We show that MIND networks had higher between subject consistency than previous approaches, greater correspondence with tract tracing data and a remarkably high level of agreement with gene co-expression networks. In the second, we show how speech transcripts can be mapped as semantic speech networks, revealing greater network fragmentation in networks from first episode psychosis patients compared to healthy control subjects (Nettekoven et al 2022, Schizophrenia Bulletin, in press).
Ultimately, combining robust estimates of structural brain connectivity with quantitative network phenotypes of behaviour could shed fresh light on the biological mechanisms underpinning mental health conditions.