University of Cambridge > > Machine Learning @ CUED > Compressed Sensing Applications in Functional Magnetic Resonance Imaging

Compressed Sensing Applications in Functional Magnetic Resonance Imaging

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Zoubin Ghahramani.

A simple, fast, and flexible 1-norm minimization method will be demonstrated to deconvolve haemodynamic response function (HRF) from functional Magnetic Resonance Imaging (fMRI) measurement. To measure individual HRF , an extra (time consuming, task limiting) calibration scan has commonly been performed before functional scans. Our method, based on HRF being sparse in a wavelet domain, instead deconvolves HRF from fMRI task data by convex optimization.

Compressed sensing conventionally means 1-norm approximation to 0-norm minimization. Advantages and limitations of the 1-norm technique and an alternative method for computing 0-norm solution will be discussed. When 1-norm method fails, we show how cardinality-constraint problems can be solved more reliably by a new method called convex iteration.

Finally, We will present a technique to design a lowpass finite impulse response (FIR) filter that minimizes the time domain impulse response peak amplitude when given a magnitude frequency response constraint (phase is unconstrained). This method uses convex optimization technique instead of combinatorial approach.

This talk is part of the Machine Learning @ CUED series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.


© 2006-2024, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity