A model-based crosstalk compensation method for simultaneous Tl-201 and Tc-99m dual isotope myocardial SPECT imaging

Clinical use of Tl-201 and Tc-99m dual isotope myocardial SPECT requires compensation for crosstalk contamination of the Tl-210 data. The most significant sources of the crosstalk are Tc-99m downscatter and Pb X-rays generated by the interaction of Tc-99m photons with the collimator. In this work we evaluated a model-based crosstalk compensation method using experimental data acquired from a torso phantom with a cardiac insert. This model-based method incorporated downscatter and Pb X-ray models to estimate and thus compensate the contamination. A cold lesion was placed in an inferior apical myocardial region of the cardiac insert. Long acquisitions were performed to obtain high-count projection data in Tl and Tc energy windows. Separately acquired Tl-201 and Tc-99m projection images were appropriately scaled and summed together to simulate separate and simultaneous acquisitions. Noisy projection images were generated by simulating Poisson noise using the high-count projection images. A total of 100 noise realizations of the projection images were used to study the image noise levels. The reconstructed Tl-201 images of simultaneous acquisition were compared to those of separate acquisition in terms of image noise and defect contrast. It was found that the model-based crosstalk compensation method substantially improved the lesion visualization and lesion contrast in the Tl images and reduced quantitative errors in polar maps. A contrast-noise curve showed that the model-based method resulted in an improved contrast-noise trade-off compared to images without crosstalk compensation. In conclusion, the results of this preliminary evaluation indicate that the model-based method is promising. However, more detailed development and evaluation of the model-based method is required to its clinical utility