Hubert Martinus Fonteijn1, Matt J. Clarkson2, Marc Modat1, Josephine Barnes2, Manja Lehmann2, Sebastien Ourselin1, Nick C. Fox2, Daniel C. Alexander1
1Computer Science, Centre
for Medical Image Computing,
This abstract introduces a novel method for studying disease progression using cross-sectional data. The model describes disease progression as a series of events and treats each data point as a snapshot of this series. We calculate the probability that an event has happened and use a MCMC algorithm to construct plausible series of events from this probability. We demonstrate our model on serial T1 MRI data from a familial Alzheimers disease cohort. We calculate regional atrophy using non-linear registration methods and show progression of atrophy on a much finer level than previous studies, confirming progression patterns from pathological studies.