Bayesian model comparison in non-linear BOLD fMRI hemodynamics

Daniel J. Jacobsen, Kristoffer H. Madsen, Lars Kai Hansen

AbstractNon-linear hemodynamic models express the BOLD signal as a non-
linear, parametric functional of the temporal sequence of local neural ac-
tivity. Several models have been proposed both for the neural activity
and the hemodynamics. We compare two such combined models: the
'original' balloon model with a square-pulse neural model [5], and an ex-
tended balloon model with a more sophisticated neural model [3]. We
learn the parameters of both models using a Bayesian approach, where
the distribution of the parameters conditioned on the data is estimated
using Markov chain Monte Carlo techniques. Using a split-half resampling
procedure [14], we compare the generalization abilities of the models as
well as their reproducibility, both for synthetic and real data, recorded
from two diŽerent visual stimulation paradigms. The results show that
the simple model is the best one for these data.
KeywordsBayes, model comparison, non-linear, BOLD fMRI, MCMC
TypeJournal paper [Submitted]
JournalNeural Computation
BibTeX data [bibtex]
IMM Group(s)Intelligent Signal Processing

Back  ::  IMM Publications