Algorithms for fast estimation of CBF & CMRO in O-15 dynamic PET: A simulation study on their performances

Positron emission tomography (PET) is capable of generating quantitative functional images of regional tissue biochemical, physiological and/or pharmacological processes. However, the computational complexity of pixel-by-pixel parameter estimation limits the clinical applications of parametric images. lit this paper, we propose two fast algorithms based on the non-uniformly sampled linear least squares (LLS) and generalized LLS (GLLS) methodologies for the estimation of cerebral blood flow (CBF) and cerebral metabolic rate of oxygen (CMRO). To validate the proposed algorithms, a simulation study was performed using the BLD software package. LLS and GLLS were compared to the non-linear least square (NLS) algorithm, using simulated data with various noise levels expected in clinical applications. The results demonstrated that the proposed algorithms provides estimates of equivalent estimation accuracy and reliability in comparison to NLS, with the computational time significantly reduced.