Fast Monte Carlo Simulation Based Joint Iterative Reconstruction for Simultaneous 99mTc/123I Brain SPECT Imaging

摘要

Simultaneous ^99mTc/¹²³I SPECT allows assessment of two functions under identical physiological conditions. Separation of these isotopes is difficult, however, because their energies are close. Most correction methods do not fully model either physical effects or patient-specific activity and attenuation distributions, and often reject scattered photons, rather than using the information they carry. We have developed a fast Monte Carlo (MC) simulation-based multiple-isotope and multiple-energy joint ordered-subset expectation maximization (JOSEM) iterative reconstruction algorithm, MC-JOSEM, that simultaneously corrects for scatter and crosstalk. We evaluated MC-JOSEM for simultaneous brain perfusion (^99mTc-HMPAO) and neurotransmission (¹²³I-altropane) SPECT. MC simulations of ^99mTc and ¹²³I studies were generated separately and then combined to mimic simultaneous ^99mTc/¹²³I SPECT. All details of photon transport through the brain, the collimator, and detector, including Compton and coherent scatter, septal penetration and backscatter from components behind the crystal, were modeled. We reconstructed images from simultaneous dual-isotope projections in two ways. First, we reconstructed the photopeak-energy-window projections (with an asymmetric energy window for ¹²³I) using the standard ordered subsets expectation maximization algorithm (AW-OSEM). Second, we reconstructed the 3-energy-window projections using a MC based iterative reconstruction that we have developed (MC-JOSEM). After fifteen iterations of reconstruction of ^99mTc images, root mean square errors (RMSE) of activity estimates in several brain structures from MC-JOSEM were reduced by 79-94% compared to AW-OSEM in the thalamus, striata, white matter, and gray matter. Similarly, RMSE of activity estimates from ¹²³I images reconstructed by fifteen iterations of MC-JOSEM were reduced by 30--58% compared to AW-OSEM in the striata and background.