Sparse-sampling is an important methodological advance in functional magnetic resonance imaging (fMRI), in which silent delays are introduced between MR volume acquisitions, allowing for the presentation of auditory stimuli without contamination by acoustic scanner noise and for overt vocal responses without motion-induced artifacts in the functional time series. As such, the sparse-sampling technique has become a mainstay of principled fMRI research into the cognitive and systems neuroscience of speech, language, hearing, and music. Despite being in use for over a decade, there has been little systematic investigation of the acquisition parameters, experimental design considerations, and statistical analysis approaches that bear on the results and interpretation of sparse-sampling fMRI experiments. In this report, we examined how design and analysis choices related to the duration of repetition time (TR) delay (an acquisition parameter), stimulation rate (an experimental design parameter), and model basis function (an analysis parameter) act independently and interactively to affect the neural activation profiles observed in fMRI. First, we conducted a series of computational simulations to explore the parameter space of sparse design and analysis with respect to these variables; second, we validated the results of these simulations in a series of sparse-sampling fMRI experiments. Overall, these experiments suggest the employment of three methodological approaches that can, in many situations, substantially improve the detection of neurophysiological response in sparse fMRI: (1) Sparse analyses should utilize a physiologically informed model that incorporates hemodynamic response convolution to reduce model error. (2) The design of sparse fMRI experiments should maintain a high rate of stimulus presentation to maximize effect size. (3) TR delays of short to intermediate length can be used between acquisitions of sparse-sampled functional image volumes to increase the number of samples and improve statistical power.
Keywords: sparse-sampling, fMRI, hemodynamic response, auditory neuroscience, HRF, speech perception, speech production
Citation: Perrachione TK and Ghosh SS (2013) Optimized design and analysis of sparse-sampling fMRI experiments. Front. Neurosci. 7:55. doi: 10.3389/fnins.2013.00055
Received: 09 December 2012; Accepted: 27 March 2013;
Published online: 18 April 2013.
Edited by:Jorge J. Riera, Florida International University, USA
Reviewed by:Jonathan E. Peelle, Washington University in St. Louis, USA
Copyright: © 2013 Perrachione and Ghosh. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.
*Correspondence: Satrajit S. Ghosh, McGovern Institute for Brain Research, Massachusetts Institute of Technology, 43 Vassar Street, Room 46-4033F, Cambridge, MA 02139, USA. e-mail: firstname.lastname@example.org
†Present address: Tyler K. Perrachione, 43 Vassar Street, 46-4037D, Cambridge, MA 02139, USA. e-mail: email@example.com