Assessment of Motion Bias on the Detection of Dopamine Response to Challenge

Abstract

11C-Raclopride (RAC) positron emission tomography (PET) is used to study dopamine response to pharmacological and behavioral challenges. Behavioral challenges produce smaller responses than pharmacological challenges and are more susceptible to sources of bias, including motion bias. The purpose of this study was to characterize the effect of motion bias within the context of a behavioral task challenge, examining the impact of different motion correction strategies, different task response magnitudes, and intra- versus interframe motion. Methods: Seventy healthy young adults were administered bolus plus constant infusion 11C-Raclopride (RAC) and imaged for 90 min on a 3-Tesla simultaneous PET/magnetic resonance (MR) scanner during which a functional MRI (fMRI) reward task experiment was conducted. Kinetic analysis was performed using an extension of the multilinear reference tissue model (MRTM), which encoded the task response as a unit step function at the start of the task (t = 40 min). The quantitative impacts of different approaches to motion correction (frame-based, reconstruction-based, none) were compared using voxel maps of change in binding potential (ΔBPND). Motion bias was compared to task effect by simulating different levels of ΔBPND (0%, 5%, 10%, 20%) in conjunction with simulating high and no motion. Intraframe motion was simulated using motion estimates derived from the simultaneously acquired MR data. The relative impact of intraframe motion was evaluated by comparing maps of bias in ΔBPND before and after applying frame-based motion correction. Results: Among the high-motion subjects, failure to perform motion correction resulted in large artifacts. Frame- and reconstruction-based approaches both corrected for motion effectively, with the former showing moderately more intense ΔBPND values (both positive and negative) in and around the striatum. At low task response magnitudes, simulations showed that motion bias can have a greater relative effect. At 5% ΔBPND, motion bias accounted for 60% of the total bias, while at 10% ΔBPND, it accounted for only 34%. Simulating high-temporal resolution motion, frame-based motion correction was shown to counteract the majority of the of the motion bias effect. The remaining bias attributable to intraframe motion accounted for only 8% of the total. Conclusion: Motion bias can have a corrupting effect on RAC studies of behavioral task challenges, particularly as the magnitude of the response decreases. Applying motion correction mitigates most of the bias, and specifically correcting for interframe motion provides the bulk of the benefit.