A generalized framework for in vivo detection of dopamine release using positron emission tomography

Voxel-level detection of task-induced striatal dopamine (DA) release in humans is achievable with dynamic PET imaging, enabling complex studies of motor, cognitive, and reward tasks. We previously introduced a data-driven methodology termed Residual Space Detection (RSD), which improved detection of low-amplitude DA release, however its applicability was limited to detection of low-amplitude and/or localized effects. Here, we generalize RSD to broader DA release scenarios by introducing a novel model-based baseline time-activity curve prediction method in combination with non-local-means clustering (RSD-Hybrid-IMRTM). In simulations, RSD-Hybrid-IMRTM outperforms our previous methodology for detecting global striatal DA release, improving absolute detection sensitivity by 18% at 5% false positive rate, while also demonstrating the ability to track the magnitude of task-induced changes in synaptic DA concentrations in a noise-robust manner. As a proof of principle, we apply RSD-Hybrid-IMRTM to healthy controls and Parkinson's disease subjects undergoing finger and foot tapping tasks. Results reveal expected group differences in parametric maps, parameter magnitudes, and functional segregation, demonstrating RSD-Hybrid-IMRTM's utility for investigating neurotransmission in human cohorts.