Dose reduction for synaptic density PET imaging in Parkinson's disease.

Area of Science:

• Neuroimaging
• Radiochemistry
• Artificial Intelligence

Background:

• Dynamic PET imaging with ¹¹C-UCB-J quantifies synaptic vesicle glycoprotein 2A (SV2A) density.
• Parkinson's disease (PD) is associated with lower SV2A density in brainstem nuclei and substantia nigra (SN).
• Reducing PET scan dose lowers radiation exposure but increases image noise, compromising quantification.

Purpose of the Study:

• To evaluate a self-supervised two-step deep image prior (TS-DIP) denoising method for SV2A PET.
• To assess the feasibility of using 1/10 of the standard PET dose for SV2A imaging in PD.
• To determine if TS-DIP can maintain quantitative accuracy and diagnostic performance at reduced doses.

Main Methods:

• Thirty healthy controls (HCs) and 30 PD patients underwent 60-minute dynamic PET scans.
• Full-count data were down-sampled to create ten independent 1/10-count dynamic datasets.
• TS-DIP was applied to denoise reduced-dose images; binding potential (BP_ND) maps were estimated and compared to full-count results.

Main Results:

• Full-count data revealed significantly lower BP_ND in SN and red nucleus (RN) in PD patients.
• 1/10-count data with TS-DIP maintained significant SN differences and improved correlation with motor severity.
• TS-DIP introduced minimal bias and restored statistical significance across noise variations.

Conclusions:

• Dynamic SV2A PET imaging at substantially reduced doses is feasible with TS-DIP.
• Advanced DL-based denoising techniques like TS-DIP support broader clinical application of low-dose SV2A PET.
• This approach holds promise for more accessible and safer SV2A PET imaging in PD research.