Chloe Hutton1, Antoine Lutti1,
Nikolaus Weiskopf1
1Wellcome Trust Centre for
Neuroimaging, UCL Institute of Neurology, University College London, London,
United Kingdom
Temporal signal-to-noise ratio (SNR) has been characterized as a function of image SNR for fMRI time series allowing physiological noise to be separated from thermal noise. This model is only valid for data acquired with a single-channel receiver coil. Here we propose an extension of this model to allow for state-of-the-art multi-channel acquisition mode, commonly used for fMRI studies. Using Monte Carlo simulations and 7T task-free fMRI, we show improved fit of the extended model for data where noise correlations are present. This extended model allows for characterization of temporal SNR and physiological noise using data acquired with multi-channel receiver coils.