Detecting and classifying neurotransmitter signals from ultra-high sensitivity PET data: the future of molecular brain imaging.

Efforts to build the next generation of brain PET scanners are underway. It is expected that a new scanner (NS) will offer anin sensitivity to counts compared to the current state-of-the-art, Siemens HRRT. Our goal was to explore the use of the anticipated increased sensitivity in combination with the linear-parametric neurotransmitter PET (lp-ntPET) model to improve detection and classification of transient dopamine (DA) signals. We simulated striatal [C]raclopride PET data to be acquired on a future NS which will offer ten times the sensitivity of the HRRT. The simulated PET curves included the effects of DA signals that varied in start-times, peak-times, and amplitudes. We assessed the detection sensitivity of lp-ntPET to various shapes of DA signal. We evaluated classification thresholds for their ability to separate 'early'- versus 'late'-peaking, and 'low'- versus 'high'-amplitude events in a 4D phantom. To further refine the characterization of DA signals, we developed a weighted k-nearest neighbors (wkNN) algorithm to incorporate information from the neighborhood around each voxel to reclassify it, with a level of certainty. Our findings indicate that the NS would expand the range of detectable neurotransmitter events to 72%, compared to the HRRT (31%). Application of wkNN augmented the detection sensitivity to DA signals in simulated NS data to 92%. This work demonstrates that the ultra-high sensitivity expected from a new generation of brain PET scanner, combined with a novel classification algorithm, will make it possible to accurately detect and classify short-lived DA signals in the brain based on their amplitude and timing.