Optimization algorithms and weighting factors for analysis of dynamic PET studies

Area of Science:

• Nuclear Medicine
• Medical Imaging
• Pharmacokinetics

Background:

• Positron emission tomography (PET) pharmacokinetic analysis relies on fitting PET data to models.
• Noisy PET data can lead to biased or unrealistic pharmacokinetic parameters.
• Optimization algorithms and weighting factors significantly influence parameter accuracy and reproducibility.

Purpose of the Study:

• To evaluate the performance of different optimization algorithms in PET pharmacokinetic analysis.
• To assess the impact of incorrect weighting factors on the accuracy and reproducibility of fitted parameters.

Main Methods:

• Compared interior-reflective Newton methods with a modified simulated annealing (SA) algorithm (basin hopping).
• Investigated the effects of various weighting factors using simulated and clinical time-activity curves (TACs).
• Utilized data from [(15)O]H(2)O, [(11)C]flumazenil, and [(11)C](R)-PK11195 studies.

Main Results:

• SA provided accurate results without needing initial parameter estimates, unlike the standard Newton method.
• Modified Newton methods and SA yielded accurate results for patient studies with high variability.
• Minor mismatches in weighting factors had little impact; large errors occurred only with significant mismatches.

Conclusions:

• Algorithm and weighting factor selection critically impacts PET pharmacokinetic analysis accuracy and precision.
• SA demonstrates superior performance, especially when initial parameter values are unknown.
• Noise reduction strategies like wavelet filtering may enhance accuracy if bias is avoided.