Costin Tanase1, Jeffrey O'Hara2, Denise Davis3, Fernando Boada3, Michael H. Buonocore4, Cameron S. Carter1
1Psychiatry & Behavioral Sciences, University of California at Davis, Sacramento, CA, United States; 2Siemens Medical Solutions; 3University of Pittsburgh, United States; 4Radiology, University of California Davis, United States
It is well accepted that the statistical assessment of fMRI data can be improved by estimating the measurement noise as well as the fMRI series fluctuations due to slow physiological processes. While recent fMRI literature has characterized a larger array of increasingly subtler physiological fluctuations, there is still an overriding assumption that the leftover variance in the data can be described by stationary Gaussian noise. By analyzing the QA data, the extracted temporal series demonstrate the presence of signal fluctuations that are non-random and consistent across multiple runs.