Modern medicine emphasizes high-value precision care. Within the context of precision health, leveraging clinical, genetic, and radiomic data in combination with applications of machine learning holds great promise in uncovering patterns within patients’ groups that may ultimately inform personalized therapeutic intervention. On the other hand, studies using positron emission tomography (PET) with radiotracers designed to bind molecular targets specifically are too small in size for this data-mining approach, and shared, substantive databases of PET data are in early stages of development. Nevertheless, the success of precision psychiatry in the years ahead is arguably limited without continued efforts using PET neuroimaging in vivo to characterize the molecular changes relevant to the pathophysiology of psychiatric conditions like psychosis. We present several new PET-based imaging agents that we designed to specifically target key proteins relevant to studying altered neuroimmunity across and within diagnostic groups of patients. First, among our radiochemistry efforts, we recently developed [18F]FNDP, which binds soluble epoxide hydrolase (sEH), an enzyme that controls the bioavailability of epoxyeicosatrienoic acids that have vasoactive and anti-inflammatory roles. sEH has been found to be elevated in post-mortem brain tissue from individuals with psychosis. Second, we developed [18F]ASEM that targets the alpha7 nicotinic acetylcholine receptor (α7-nAChR), which has a role in anti-inflammatory cholinergic signaling. Building on postmortem evidence of low hippocampal expression of the α7-nAChR in patients with non-affective psychosis (NP), we used [18F]ASEM PET to test whether individuals with NP have lower [18F]ASEM binding compared to healthy individuals or individuals with affective psychotic disorder (AP). Finally, we recently developed [11C]CMPPF for imaging the microglia-specific target, colony stimulating factor 1 receptor (CSF-1R), due to evidence that activated microglia may be mechanistically linked to psychosis in some individuals. After bolus injection of [18F]FNDP, high levels of brain radioactivity were observed in healthy participants, with peak gray matter activity within 10 minutes post injection and steady decline over the remaining scan. [18F]FNDP total distribution volume (VT) was well-estimated in several brain regions from 90-minute data using the 2-tissue compartment model with metabolite-corrected arterial input function, supporting early efforts to use [18F]FNDP PET in clinical research. Using [18F]ASEM PET, VT was lower in individuals with NP (14.1 ± 0.9) compared to healthy controls (19.6 ± 2.5, P < 0.001) or compared to those with AP (17.6 ± 2.2, P = 0.04). Among patients, higher [18F]ASEM VT was associated with better processing speed and verbal memory after adjusting for age. Finally, [11C]CMPPF with PET in living baboon brain shows promising, high uptake and specific binding, as well as higher VT after systemic administration of lipopolysaccharide to activate microglia. Movement of this radiotracer into human neuroimaging studies is underway toward the ultimate goal of testing whether [11C]CMPPF PET supports the hypothesized subgroup of individuals with psychosis who have aberrantly high microglial response. Use of PET-based imaging even in smaller-scale studies may facilitate clinically meaningful subtyping of individuals with aberrant neuroimmune response among those with psychosis. Future work should pursue evaluation of approaches such as a generative adversarial network to augment what can be discovered from relatively small PET samples, and assessment of clinical decision-making value from use of a growing number of radiotracers for PET-based molecular phenotyping of subpopulations of patients.