[PRN_141209001]Linear trend of resting-state fMRI time series

Linear trend of resting-state fMRI time series

Xin-Di Wang^{a, b}, Chao-Gan Yan^c,d,*, Yu-Feng Zang^{e, f,}*

^aState Key Laboratory of Cognitive Neuroscience and Learning & IDG/McGovern Institute for Brain Research, Beijing Normal University, Beijing, 100875, China

^bCenter for Collaboration and Innovation in Brain and Learning Sciences, Beijing Normal University, Beijing, 100875, China

^cThe Nathan Kline Institute for Psychiatric Research, 140 Old Orangeburg Road, Orangeburg, NY, 10962, USA

^dDepartment of Child and Adolescent Psychiatry, New York University Langone Medical Center, New York, NY, 10016, USA

^eCenter for Cognition and Brain Disorders, Hangzhou Normal University, Hangzhou, 311121, China

^fZhejiang Key Laboratory for Research in Assessment of Cognitive Impairments, Hangzhou, 311121, China

^*Corresponding:

Chao-Gan Yan, The Nathan Kline Institute for Psychiatric Research, 140 Old Orangeburg Road, Orangeburg, NY, USA.

Email: ycg.yan@gmail.com

Yu-Feng Zang, Center for Cognition and Brain Disorders, Hangzhou Normal University, Hangzhou, 310015, China.

email: zangyf@gmail.com

Abstract: Although linear trend removing has often been implemented as a routine preprocessing step in resting-state functional magnetic resonance imaging (RS-fMRI) data analysis, the spatial distribution of the magnitude of linear trend is still unclear. Further, it is interesting whether there will be any difference of the linear trend magnitude between different resting-states. For the first aim, we analyzed 5 RS-fMRI datasets from 5 different scanners (namely Beijing-Simens-3T, Cambridge-Siemens-3T, CCBD-GE750-3T, Milwaukee-GE-3T, and Oulu-GE-1.5T). One-sample t-tests on the regression coefficient (i.e., the magnitude of linear trend) were performed for each datasets. For the second aim, we used 2 datasets in each of which different states were compared, one containing eyes-open resting-state (EO-RS) vs. eyes-closed resting-state (EC-RS) and the other containing two steady-state tasks, i.e., real-time finger force feedback (RT-FFF) and sham finger force feedback (S-FFF) tasks. Paired t-tests were performed between EO-RS and EC-RS, and between RT-FFF and S-FFF. One-sample t-tests showed that the spatial pattern of linear trend of RS-fMRI time series were quite different between different manufactures. The 3T SIEMENS scanners showed positive linear trend in almost all part of the brain, while GE scanners showed primarily negative linear trend in most part of the brain. Paired t-tests showed some differences between paired conditions; differences between EO-RS and EC-RS were mainly in cuneus and eyeballs, and differences between RT-FFF and S-FFF were found in the thalamus, anterior cingulate gyrus, and right sensorimotor cortex. The current study indicated that, while the manufacturer-dependent linear trend of RS-fMRI time series were mostly scanner-related noise, the linear trend may also be physiological noise (eyeballs) or even physiologically meaningful, especially during steady-state tasks.