Cerebral Palsy (CP) is the most common motor disability in children, characterised by impaired movement, muscle coordination, and maintenance of posture. Early diagnosis of CP is essential, as early treatment and intervention can drastically improve the outcomes. A recent study has shown that the use of Magnetic Resonance Imaging (MRI) can help diagnose CP as early as 5 months, however, it requires specialist detection of brain malformation. Thus, an automated pipeline that can detect malformation from an MRI scan, i.e. by measuring specific tissues’ volumes and comparing them to the expected development, has the potential to aid the early detection of CP. However, while well-established neuroanalysis tools exist for adults, their application in infants (the period between 2 months and 2 years of age) is unfeasible because of the vast difference in brain size and the changes in the appearance of white and grey matter. Within this project, we aim to improve infant neuroimaging by developing a new state-of-the-art (SOTA) pipeline, to quantify brain development between 3 months to 2 years, a period of vast change. The project is a collaborative effort with Lilla Zöllei, the developer of the current SOTA infant neuroimage analysis tool, which we aim to improve upon by using deep-learning algorithms.

Once we have developed an algorithm to analyse the infant brain, we will explore the differences between CP and expected brain developments. To facilitate this investigation, I will use data collected as part of the NIBS-CP and CP-EDIT: two large studies that are running in parallel to follow the development of (up-to) 200 subjects; 50-75 of whom are high-risk of CP and the remaining healthy controls. Each recruit will have a series of longitudinal MRI scans at 3-9 months, 12 months and 24 months, paired with developmental outcomes. Thus, there is the potential to detect early CP biomarkers from 3 months. This rich data can then be used to build normative models of a typically developing Danish infant population and investigate whether we can identify predictive features of CP and motor deficits in the CP population. We hypothesise that accurately quantified brain structures can facilitate earlier diagnosis of CP and predict developmental outcomes, potentially leading to earlier intervention and thus better outcomes. Further, a normative model of typically Danish child development may promote the early diagnosis of other developmental disorders.