Abstract #4134
Simultaneously resolve haemodynamic response function and activation response by rank-constrained optimization
Christine Law 1,2
1
Stanford University, Stanford, CA, United
States,
2
Nuffield
Department of Clinical Neurosciences, University of
Oxford, Oxford, Oxfordshire, United Kingdom
A novel technique to concurrently resolve HRF and
quantify fMRI activation response is presented. The
problem formulation involves a rank 1 constraint
(nonconvex), which makes it very difficult to solve. We
were able to compute rank minimization by reformulating
the nonconvex problem into a sequence of convex
problems. We also developed an acceleration method such
that computation time (global convergence) is
significantly improved. We achieved over 99.9% accuracy
on both HRF and stimulation response in simulations
using parameters typical to fMRI studies. Our technique
can benefit research involving a population having HRF
that deviates from the canonical assumption.
This abstract and the presentation materials are available to members only;
a login is required.
Join Here
