Skip to Main Content
Osteomyelitis (OM), a bone-marrow infection, is best diagnosed by simultaneous In-111 white blood cell and Tc-99m sulfur colloid imaging. The diagnosis has conventionally been based on a visual assessment of images, where a region of increased In-111 activity, spatially concordant with a region of decreased Tc-99m activity, indicates possible OM. To obtain more accurate and precise tomographic images of both radionuclides, we modified a previously developed Monte Carlo (MC)-based scatter and crosstalk estimation procedure based on a joint ordered-subsets expectation-maximization (MC-JOSEM) reconstruction algorithm, and compared its performance to that of a triple-energy-window-based (TEW-OSEM) reconstruction approach. A numerical phantom with eight simulated bone-marrow infection regions was used to evaluate the performance of both methods in simultaneous In-111/Tc-99m imaging. After using our SPECT MC program to simulate essentially noise-free data for both tracers, eight independent projections of Poisson-noise deviates were generated. Volumes of interest (VOI) were drawn in five normal-marrow and eight infected-marrow locations to assess the performance of both methods, with respect to that achievable using primary photons only from both nuclides. Overall, MC-JOSEM (χ2=1.7) outperformed TEW-OSEM (χ2=7.4). After one iteration, which provided the lowest mean-squared error of activity estimates, the average relative bias ± precision of reconstructed Tc-99m activity concentration in regions of infection decreased from 81.6±17.7% for images not compensated for scatter (NC) to 11.1±12.3% (MC) vs. 16.9±11.0% (TEW) and, in normal-bone regions, decreased from 34.4±22.5% (NC) to -3.9±10.8% (MC) vs. 4.1±11.5% (TEW). For In-111, the relative accuracy ± precision in regions of infection improved from 11.6±14.1% (NC) to 0.6±9.9 % (MC) vs. 15.5±8.5% (TE- ) and, in normal-bone regions, decreased from 75.8±31.2% (NC) to 34.8±26.4% (MC) vs. 60.6±21.2% (TEW).