Cross-scanner validation and time-stability analysis of STARE, a blood-free quantification tool for brain imaging with irreversible PET tracers

OBJECTIVE: "Gold standard" blood-based quantification of dynamic 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography (PET) data has limited practical clinical applications due to cost and complexity of data collection and analysis. We previously presented a blood-free quantification alternative, STARE (Source-to-Target Automatic Rotating Estimation), that was validated on 18F-FDG data acquired on an ECAT EXACT HR+ scanner. Here, we extend that initial work by applying STARE to within-subject data acquired with both a Siemens Biograph mCT scanner and a portable Brain Biosciences CerePET scanner. Approach: Performance was assessed by comparing regional net influx rates (Ki) estimated using STARE and the standard blood-based Patlak approach. Twenty participants underwent 60-minute brain 18F-FDG scans, on two different days, once in each scanner. The time-stability of both STARE- and Patlak-based Ki estimates was evaluated by applying each method to the first 20 (STARE only), 30, 40, and 50 minutes of data. Main results: STARE demonstrated high correlation with Patlak Ki estimates, particularly in the Biograph mCT (r = 0.82-0.93 (within-region)), with lower correlation in the CerePET (r = 0.27-0.62 (within-region)). In the Biograph dataset, STARE showed small deviations from the 60-min estimates across all evaluated durations (20 min and above), whereas deviations increased with shorter durations in CerePET data. The Patlak approach showed smaller changes across abbreviated scan durations in both datasets. Significance: STARE shows potential as a noninvasive alternative to traditional blood-based quantification of dynamic 18F-FDG PET brain data, facilitating shorter, blood-free acquisition. This advancement could make dynamic 18F-FDG PET brain imaging more accessible and comfortable for patients, potentially facilitating broader clinical adoption. .