Characterizing acoustic imaging noise in auditory cortex event-related functional magnetic resonance imaging
Abstract
Acoustic imaging noise (AIN) produced during functional magnetic resonance imaging (fMRI) studies can hinder auditory fMRI research analysis by altering the properties of the acquired time-series data. AIN can be especially confounding when estimating the time-course of the hemodynamic response (HDR) in event-related fMRI (ER-fMRI) experiments. Therefore, given the desire to obtain the most accurate estimates of the HDR using ER-fMRI, analysis techniques need to be developed that will account for measurable properties of the cortical response to AIN. Three projects were completed that will help achieve this goal, and possible mechanisms to compensate for AIN are presented. The first and second projects characterized and modeled the amplitude, spatial extent, and linearity of the HDR to AIN associated with echo-planar acquisition. Results from these studies indicate that the estimated HDRs in auditory cortex to 1-, 5-, and 15-ping stimuli, as well as two 5-ping stimuli presented two seconds apart (5-5-ping), are comparable in amplitude and duration to a typical stimulus-induced HDR in auditory cortex. Non-linearity of the HDR to increasing amounts of AIN (1-, 5-, and 15-ping) is observed, but there is evidence of linearity of the HDRs when the stimuli are sufficiently spaced, based on the comparison between the HDRs to the 5-ping and 5-5-ping stimuli. The third project quantified the dependence of the HDR on AIN and estimated how the effect of AIN on ER-fMRI experiments is dependent on repetition time (TR) and the inter-stimulus interval (ISI). From this study, it is evident that the effect of AIN is dependent on the TR parameter. Extent of dependence on ISI was not able to be determined, although it was found that responses were similar for ISI = 12 and 18 s. Finally, possible mechanisms to compensate for AIN are presented, illustrating how results from the three projects could be used to either refine the existing general linear models or establish new models that could account for AIN.
Degree
Ph.D.
Advisors
Talavage, Purdue University.
Subject Area
Electrical engineering|Biomedical research|Neurology
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.