The goal of this work is to develop reconstruction and acquisition techniques that allow simultaneous acquisition of projections from two radionuclides and provide reconstructed SPECT images representing the activity distribution of each. The research currently focuses on two clinical applications: rest/stress myocardial perfusion imaging using Tc-99m and Tl-201 labeled agents and brain imaging using Tc-99m and I-123 labeled isotopes.

Dual isotope imaging results in significant crosstalk. Figure 1 below shows the energy spectrum from Tc-99m and Tl-201 distributions in a body-like object. Note that Tc-99m photons scattered in the body and lead x-rays created in the collimator will be detected in the Tl-201 energy window. Figure 2 illustrates the energy spectrum for I-123 and Tc-99m brain imaging. In this case the photopeak energies are close together, resulting in detection of both scattered and unscattered photons from each isotope in the photopeak energy window of the other isotope. In addition, high energy (> 300 keV) I-123 photons interact in the patient, collimator, and detector and are detected in both the I and Tc energy windows.

Figure 1. Sample energy spectra of Tl-201 and Tc-99m

Figure 2. Sample energy spectra of Tc-99m and I-123.

If SPECT images are reconstructed from the crosstalk-contaminated projection data discussed above results, the reconstructed images will be degraded. Figure 3 shows sample images of a heart phantom reconstructed from the Tl-energy window projection data when Tc is absent and present in the phantom. A defect is present in the lateral myocardial wall. Note the reduced contrast with Tc present. The crosstalk also affects the quantitative accuracy of the images. Figure 4 shows the decrease in quantitative accuracy in quantifying the activity in the striatum for separate acquisition of brain images using a Tc-labeled agent with and without I-123 present. The figure also shows the improved of the accuracy after crosstalk compensation.

Figure 3. Illustration of effect of crosstalk on image quality.

Figure 4. Illustration of degradation in image quantitative accuracy due to crosstalk for I-123/Tc-99m brain imaging. In general, crosstalk causes overestimation in absolute quantitation. The improvement after model-based crosstalk compensation (MBCC) is also shown in the figure.

 We have developed methods to model the crosstalk, include it an an iterative reconstruction algorithm, and thus compensate for it. The process for simultaneous Tc-99m/Tl-201 myocardial perfusion imaging is illustrated in Figure 5. For I-123/Tc-99m imaging, simultaneous reconstruction of the activity estimates from both isotopes is performed using crosstalk modeling, as illustrated in Figure 6.

Figure 4. Compensation process for simultaneous Tc-99m/Tl-201 imaging. A sequential compensation method is used because contamination of the Tc data is relatively small.

Figure 5. Illustration of a simultaneous reconstruction process used for crosstalk compensation in I-123/Tc-99m brain imaging.

The crosstalk compensation methods result in both improved image quality and quantitative accuracy. Figure 6 shows Tl-201 heart phantom images from separate acquisition (no contamination) and from simultaneous acquisition with and without compensation. Note that with compensation the images are essentially identical to those from separate acquisition. Also note that the results for the model-based method are very close to those when we know the true compensation. The method also results in improved quantitative accuracy. Figure 7 illustrates the improvement in quantitative accuracy for estimation of the Tc-99m and I-123 binding potential in the striatum for simultaneous acquisition with and without crosstalk compensation.

Figure 6. Illustration of efficacy of crosstalk compensation. The images with compensation are virtually identical to those from Separate acquisition.

Figure 7. Illustration of improvement of quantitative accuracy for simultaneous striatum imaging. The figure shows the improvement in binding potential (BP) estimation. BP was calculated as the activity concentration ratio between striatum and background.

Eric C. Frey

Yong Du

Bin He

Martin J. Stumpf

Stacia A. Sawyer

This work is funded by the NIBIB under grant R01EB000288, “Simultaneous Dual Isotope Imaging with Crosstalk Correction”.