Activity estimation in small volumes with non-uniform radiotracer uptake using a local projection-based fitting approach

摘要

We have previously evaluated a local projection-based approach that provides robust estimates of activity concentration in small volumes-of-interest (VOI) affected by partial volume and tissue crosstalk in clinical SPECT imaging. The approach requires local segmentation of functionally distinct tissues within a VOI from a registered, high-resolution anatomical image of the object. Measured projection data are fitted to a statistical model of segmented-tissue projections. The resulting linear equations are solved to recover corrected values of tissue-activity concentration. In this work, we extended the approach to incorporate models of non-uniform radiotracer uptake into the fitting procedure. We evaluated the modified method using 25 independent noise realizations of simulated torso phantoms, each containing 20 identical, spherical “tumors” within a homogenous activity background. Radially varying quadratic functions were used to simulate two degrees (50% and 90%) of reduced central uptake inside the tumors (e.g., from necrosis), with an average integrated tumor-to-background concentration ratio of 7∶1. Tumor-activity estimates were obtained by fitting projection data to models that assumed either uniform tracer uptake (UF) or radially varying non-uniform uptake (NUF). The sensitivity of the approach to registration errors was investigated by simulating a 1-pixel misspecification of the locations of the VOI. The NUF approach achieved better than 1% bias and 12% precision for total tumor-activity estimates. Increasing the degree of central count loss from 50% to 90% did not significantly affect the NUF bias. In comparison, the UF method yielded 16% bias and 1% precision (48% bias, 5% precision) for 50% (90%) central count loss. The simulated registration error did not affect the UF estimates, but degraded the accuracy (and precision) of NUF estimates to 11% (and 13%). Despite slightly inferior precision, NUF may permit improved assessment of- inhomogeneous tumor uptake.