Meeting Banner
Abstract #1712

Deconvolving Haemodynamic Response Function in FMRI Under High Noise by Compressive Sampling

Christine Law1, Gary Glover1

1Stanford University, Stanford, CA, USA

A simple technique to deconvolve haemodynamic response function (HRF) from fMRI data using 1-norm minimization is introduced. The true HRF is typically sparse after wavelet transform, but we find the proposed technique to be robust w.r.t relative sparsity. HRF is deconvolved via convex optimization which has the flexibility to impose local HRF monotonicity and smoothness in the time domain. Real fMRI data under low SNR (-10dB) confirms reliability of this technique.