Title :
Out-of-plane photon compensation for 3-D SPECT image reconstruction with generalized matrix inverses
Author_Institution :
Dept. of Biomed. Eng., Duke Univ., Durham, NC, USA
Abstract :
A computationally efficient 3-D image reconstruction method which compensates for detected out-of-plane photons has been developed for SPECT image reconstruction with generalized matrix inverses (GMI). Fully 3-D image reconstruction is approximated by a series of coupled 2-D image reconstructions for projection data acquired with parallel hole collimators, significantly reducing computer memory requirements. In this method, projection data are compensated for detected scattered photons using dual energy window scatter subtraction (Step 1). An initial source activity estimate in each transverse plane is then made using line source response functions in the system matrix (Step 2). With these activity estimates the contributions of out-of-plane unscattered photons are modeled and subtracted from the projection data, and an updated source activity estimate is computed (Step 3). For noise-free projection data from a Monte Carlo simulated myocardial perfusion study, lesion contrast increases and activity spillover from the myocardium into the adjacent cardiac blood pool is reduced. For projection data with Poisson noise, activity estimates from Step 3 are inferior to those from Step 2. One reason is that the relative noise level of the projection data is substantially increased by subtracting the estimated out-of-plane contribution. The Step 3 activity estimation is also sensitive to the source activity distribution used for out-of-plane unscattered photon compensation. Though the out-of-plane compensation technique of Step 3 provides some benefit for noise-free projection data, this study suggests that it may not be well-suited for practical application to count-limited clinical SPECT studies.<>
Keywords :
image reconstruction; medical image processing; single photon emission computed tomography; 3D SPECT image reconstruction; Monte Carlo simulated myocardial perfusion study; Poisson noise; activity spillover; coupled 2D image reconstructions; dual energy window scatter subtraction; generalized matrix inverses; lesion contrast increases; medical diagnostic imaging; noise-free projection data; nuclear medicine; out-of-plane compensation technique; out-of-plane photon compensation; out-of-plane unscattered photons; scattered photons detection; Collimators; Concurrent computing; Electromagnetic scattering; Image reconstruction; Monte Carlo methods; Myocardium; Noise reduction; Optical computing; Particle scattering; Single photon emission computed tomography;
Journal_Title :
Nuclear Science, IEEE Transactions on