T1/2w ratio similarity networks: biological validation and age-related change in the common marmoset

Brisbane Convention & Exhibition Centre 

Understanding developmental changes in brain structure is a key goal of psychiatry as the peak incidence of many mental disorders falls in adolescence. MRI enables non-invasive imaging of brain structural networks, where regions are modelled as nodes and connections as edges [1]. Structural similarity networks, based on statistical relationships between regional morphological features [2], have revealed distinct developmental processes within paralimbic and isocortical regions [3]. Perturbation to these processes may underlie globally reduced structural similarity seen in psychosis [4].

Morphometric Inverse Divergence (MIND), a novel method for estimating structural similarity from MRI, captures known aspects of biology and outperforms DTI in age prediction tasks [5]. However, similarity networks generated from multiple features can be hard to interpret. Here, we generate MIND similarity networks using T1/2w ratio, interpretable as a measure of myelin [6], and assess 1) whether networks are biologically valid and 2) what age-related changes are present in these networks. We use the common marmoset as an animal model due to an abundance of openly accessible biological data against which to validate T1/2w networks.