Meeting Banner
Abstract #2867

Columnar Organization or Noise? Optimization and Validation of FMRI at the Resolution of Columns

Denis Chaimow1, 2, Kamil Ugurbil1, Amir Shmuel, 13

1Center for Magnetic Resonance Research, University of Minnesota, Minneapolis, MN, United States; 2Max Planck Institute for Biological Cybernetics, Tbingen, Germany; 3Montreal Neurological Institute, McGill University, Montreal, QC, Canada

The ability of high-field fMRI to truly resolve columnar organization depends on the spatial scale of the pattern, the BOLD point spread, the voxel size and the noise level. To better understand the role of each factor, and to guide the selection of optimal parameters, we developed a mathematical model of imaging cortical columns. We quantified the expected differential functional contrast relative to noise and the expected similarity between the imaged pattern and the true columnar organization as a function of parameters of interest. We found that the voxel width that optimizes differential contrast is larger than the one that optimizes accuracy. Furthermore, we propose a method for confirming that an imaged pattern reflects the true columnar organization.

Keywords

abruptly accuracy accurately acquisition address adequate advent affected allows applying assuming attempted become better bingen biological blood bold capacity certain columnar columns confirm considering contrast contributions correlation cortical cybernetics cycle decreases degree depend dependent depends detect developed differential enough evaluated expected falls field firstly frequency function functional generated generic gradually guide hand highest increasing indicate institute interplay inverse investigated irregularity isotropic knowledge known larger length limit maps mathematical maximized measures model modeled moderate mostly neurological neuronal noise obvious optimal optimization optimizes organization organizations original oxygenation pattern patterns peak power quantified question reconstruct reconstructed reflects regularity relying resolution resolvability resolving respectively response role sampled sampling scale secondly selection several similarity simulated simulations spatial spectrum spin spread step sufficiently supported third true underlying understand unimodal unknown validation varying visible volumes whether wide width