Examining the Whiteness of fMRI Noise

O. Friman, I. Morocz, C.-F. Westin
ISMRM'05, Miami, USA
Proceedings of the ISMRM Annual Meeting (ISMRM'05)
May, 2005

Abstract

The temporal autocorrelation in fMRI time series is a source of much concern in the neuroimaging community. The serial dependence between samples stems from deterministic signals, such as scanner drifts and cardiac and respiratory cycles, about which we have very little prior knowledge and therefore may have difficulties modeling. In this short exposition, using over 1,000 fMRI data sets, we investigate and demonstrate how the model of irrelevant deterministic components in fMRI time series influences the residual noise structure. In particular, we show that given an adequate model for the nuisance deterministic signals, the noise in a vast majority of fMRI time series can be considered as white, i.e. temporally uncorrelated.

Reference

Friman O, Morocz I, Westin CF. Examining the whiteness of fMRI noise. In Proceedings of the ISMRM Annual Meeting (ISMRM'05). Miami, USA, 2005;.

Bibtex entry

@InProceedings{frimanISMRM05,
  author         = {O. Friman and I. Morocz and C.-F. Westin},                 
  title          = {Examining the Whiteness of {fMRI} Noise},                  
  booktitle      = {Proceedings of the {ISMRM} Annual Meeting (ISMRM'05)},     
  year           = {2005},                                                     
  address        = {Miami, USA},                                               
  month          = {May}
}

Grants

NIH P41-RR13218 (NAC)

Research area

fMRI