Computer Assisted Screening for Cortical Alterations in Development (CASCADE)

Neurocognitive conditions such as epilepsy, learning impairments, autistic spectrum conditions (ASC) and ADHD impact 3-4% of all children in the UK. However, often these conditions are not diagnosed until symptoms first present during mid-childhood, at which point the window for most effective intervention may have closed. This has both societal and personal consequences, since considering ASC alone, less than 22% of affected individuals currently find full time employment, with costs of care and lost earnings exceeding £32bn pa. The first objective of this proposal is to develop a suite of tools that support precise, early phenotyping of these conditions from fetal magnetic resonance imaging (MRI), with view to improving clinical understanding of how (and when) the brains of these individuals deviate from expected patterns of development, in order to inform diagnosis and improve targeting of early infant support. This is motivated by a growing body of evidence that suggests that atypical neurodevelopment starts in the womb[1][2][3]. It also follows a trend towards increased clinical use of MRI, in cases of suspected genetic/clinical risk, or from concerns raised during routine ultrasound. This is because MRI generates clear and detailed images that improve detection and characterisation of alterations, making it vital when guiding parents towards the need for postnatal support/monitoring. Despite these benefits, early imaging-biomarkers of neurodivergent development remain elusive. One key reason for this is that these conditions impact the cortex, or layers of tissue at the surface of the brain, which fold during neurodevelopment in order to fit within the skull. This biomechanical process is sensitive to any small changes in initial conditions and/or the rate of cell growth. As a result every human’s brain folds differently, and this makes it very challenging for clinicians to determine whether any unusual pattern of folding falls within the range of normal variation, or reflects early signs of cortical malformation. Our team brings together world-leading expertise in fetal MRI reconstruction and cortical modelling, with clinicians who have pioneered the use of fetal MRI for clinical reporting, and longitudinal research studies. Collectively we have accumulated >3000 scans, representing a range of clinical/reference populations and acquired across a range of acquisition protocols. This puts us in a unique position to be able to build models which disentangle natural cortical shape variability and the effect of scanning protocols, from neurobiological phenotypes of interest. Moreover, our methods will be informed by current understanding of the cellular biology underpinning cortical development, to annotate the sites of major cortical folds as they emerge, while sampling corresponding locations in the transient regions of the subplate and ventricular zones, from which cortical neurons emerge and migrate. From this we can then extract metrics of tissue maturation from these locations, harmonised across clinical and reference populations, to build normative reference charts from which early markers of atypical cortical development may be detected as outliers. Our vision is that these tools be failure-proof and simple-to-use - enabling users to reconstruct anatomically-precise surface-models of the fetal cortex and transient layers, with all emerging folds automatically labelled. As a result, our secondary objective is to trial use of these visualisation tools as supplements to clinical reporting - reducing the time needed for neurologists to read images, while making it easier to appraise the entire shape of the fetal cortex from a single holistic view.