Differentiating BOLD and non-BOLD signals in fMRI time series from anesthetized rats using multi-echo EPI at 11.7T

Abstract

The study of spontaneous brain activity using fMRI is central to mapping brain networks. However, current fMRI methodology has limitations in the study of small animal brain organization using ultra-high field fMRI experiments, as imaging artifacts are difficult to control and the relationship between classical neuroanatomy and spontaneous functional BOLD activity is not fully established. Challenges are especially prevalent during the fMRI study of individual rodent brains, which could be instrumental to studies of disease progression and pharmacology. A recent advance in fMRI methodology enables unbiased, accurate, and comprehensive identification of functional BOLD signals by interfacing multi-echo (ME) fMRI acquisition, NMR signal decay analysis, and independent components analysis (ICA), in a procedure called ME-ICA. Here we present a pilot study on the suitability of ME-ICA for ultra high field animal fMRI studies of spontaneous brain activity under anesthesia. ME-ICA applied to 11.7T fMRI data of rats first showed robust performance in automatic high dimensionality estimation and ICA decomposition, similar to that previously reported for 3.0T human data. ME sequence optimization for 11.7T indicated that 3 echoes, 0.5mm isotropic voxel size and TR=3s was adequate for sensitive and specific BOLD signal acquisition. Next, in seeking optimal inhaled isoflurane anesthesia dosage, we report that progressive increase in anesthesia goes with concomitant decrease in statistical complexity of “global” functional activity, as measured by the number of BOLD components, or degrees of freedom (DOF). Finally, BOLD functional connectivity maps for individual rodents at the component level show that spontaneous BOLD activity follows classical neuroanatomy, and seed-based analysis shows plausible cortical–cortical and cortical–subcortical functional interactions.