24 Generalisable MRI-based normative modelling for detecting age-inappropriate neurodegeneration

a:2:{s:4:"lang";s:2:"en";s:7:"content";s:1598:" Determining whether MRI brain scans demonstrate atrophy that is beyond “normal for age” is challenging. Automated measurements of structural MRI have shown promise as biomarkers of neurodegeneration, yet interpretable reference standards are lacking. Normative modelling, enabling standardised “brain charts”, address this challenge by generating individualised age- and sex- adjusted centile scores derived from large, aggregated MRI datasets. Using normative data from 56,173 participants, we have developed regional cortical thickness and hippocampal volume brain charts. We applied this to a range of cohorts including individuals with evidence of Alzheimer’s disease (AD) at post-mortem (n=351); amyloid-β-PET-positive individuals with AD dementia (n=39) and mild cognitive impairment (n=71); and individuals with frontotemporal lobar degeneration (FTLD) (n=113). In AD, regional centile scores from brain charting strongly predicted cognitive performance and reliably discriminated disease groups from propensity matched cognitively normal controls. The strongest classification performance was seen in individuals with AD dementia (AUC=0.96). Distinct patterns of age-inappropriate cortical atrophy were also evident in different phenotypes of FTLD. We also demonstrate its application to individual cases from a cohort of people living with dementia. Regional cortical thickness and hippocampal volume brain charts objectively identify atrophy in neurodegenerative diseases and can be applied at the individual level. thomasparker{at}nhs.net ";}