Purchase this article with an account.
Edgar DeYoe, Ryan Raut, David Ritchie, Jed Mathis; Concordance of Resting-State vs Task-Based FMRI Maps of Human Visual Cortex. Journal of Vision 2016;16(12):881. doi: https://doi.org/10.1167/16.12.881.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Resting-state FMRI (rsFMRI) requires minimal subject participation yet provides brain maps of vision- and other function-specific networks making it highly advantageous for clinical brain mapping. However, rsFMRI and conventional task-FMRI (taFMRI) may not reflect identical neural mechanisms. To compare rsFMRI and taFMRI maps of visual cortex on a voxel-by-voxel basis, three healthy individuals and 5 brain tumor patients underwent taFMRI vision mapping with an expanding checkered annulus extending to 20o eccentricity plus rsFMRI with eyes closed. TaFMRI data were subjected to both independent components analysis (ICA) and correlation analysis (COR) whereas rsFMRI data were processed with ICA. Activation patterns were compared across a wide range of statistical threshold settings. For each voxel, the presence (+) or absence (-) of rsFMRI and taFMRI activation was scored as one of 4 logical combinations (+/+,+/-,-/+,-/-). Brain maps of the voxel classifications revealed regions of high predictive validity (+/+ and -/-) versus zones of mismatch (+/- and -/+). Predictive validity was threshold dependent, but for each subject a single pair of taFMRI and rsfMRI thresholds was found that minimized this dependence and was unbiased. At such thresholds, the average predictive validity was 0.62 for taFMRI-COR vs rsfMRI-ICA, and 0.59 for taFMRI-ICA vs rsfMRI-ICA, similarly for both healthy subjects and tumor patients. Mismatches were non-random, often located at the edges of the fMRI patterns partly reflecting the more limited retinotopic extent of the taFMRI activation compared to rsFMRI. TaFMRI maps computed from the same dataset with ICA vs COR resulted in higher, but not perfect, predictive validity (0.75) suggesting that analysis method can be a significant source of variability. Conclusion: taFMRI and rsFMRI maps can be similar but the degree of concordance is threshold and analysis dependent. rsFMRI may provide more complete maps including regions not activated by taFMRI.
Meeting abstract presented at VSS 2016
This PDF is available to Subscribers Only