Energy Window and Contrast Optimization for Single-photon Emission Computed Tomography Bremsstrahlung Imaging with Yttrium-90

Introduction

In gamma camera imaging, the acquisition energy window is centered around photopeak to detect majority of primary photons. However, for yttrium-90 (Y-90) bremsstrahlung, as the acquired spectrum is complex and continuous, the choice of acquisition energy windows is one of the most challenging topics in nuclear medicine.[1] Several works have been performed in the objective to optimize bremsstrahlung imaging.[23456] However, no study has evaluated the image quality and accurate activity quantification for Y-90 bremsstrahlung in terms of contrast and contrast to noise ratio and also geometric, penetration, and scatter components. In this study, a Monte Carlo simulation SIMIND code[7] was used to investigate the effects of the energy windows, using a high-energy (HE) collimator on the image contrast and signal to noise ratio (CNR), in order to optimize the Y-90 bremsstrahlung single-photon emission computed tomography (SPECT) imaging. The simulations were set up in such a way that it provides geometric, penetration, and scatter components to a separate file. At the end of simulations, binary images were imported in ImageJ software (Version 1.51) National Institutes of Health and the Laboratory for Optical and Computational Instrumentation (LOCI, University of Wisconsin).[8]

Materials and Methods

We simulated the Siemens Medical System Symbia equipped with a HE collimator and with detector having the following characteristics: 0.95 cm NaI (Tl) crystal thickness, 50 cm × 50 cm of area, intrinsic spatial resolution of 0.360 cm, and energy resolution of 10% at 140 keV. The collimator data used during the simulation are given in Table 1.

Bremsstrahlung energy spectra were generated with SIMIND Monte Carlo code (version 6.1) by simulating six spheres of different sizes filled with Y-90 and located inside water cylindrical phantom (L: 10 cm, rayon 1: 11 cm, and rayon 2: 12 cm). The inner diameters of the six spheres used are: 3.7, 2.8, 2.2, 1.7, 1.3, and 1 mm. The activity concentration for the six spheres was 3.374 MBq/mL. The phantom was positioned at 15 cm from the detector surface. The projections were generated in matrices of 256 × 256 pixels, 0.24 cm pixel size, 128 views, and 360° clockwise gantry rotation. The simulation is done, starting with large acquisition windows [Table 2]. In order to refine the results, narrower acquisition windows were considered [Table 3].

Contrast and CNR were calculated by the following formulas:

C_S = S_C/V_S: Number of counts in the spheres per voxel

SC: Total counts in the spheres

VS: Sphere volume in number of voxels

C_b = N_C/V_B: Number of counts in the background compartment per voxel

N_C: Total counts in the background compartment

V_B: Background volume in number of voxels

σ_s:The variance in sphere

σ_b: The variance in background.

Results

Figure 1 shows the energy spectrum as a function of energy for Y-90.

Close

Figure 1

Monte Carlo simulated spectrum for yttrium-90

Export to PPT

As shown in Figure 2, the geometric component rehearses a high value in 135 keV for 1^st acquisition and 130 keV in the second. Whereas, the scatter and penetration components are small at this energy in both cases.

Close

Figure 2

Contributions of geometric, penetration, scatter, and X-rays components with energy windows

Export to PPT

Figure 3 shows the effects of the energy window on the image contrast of the hot spheres with HE collimator. In comparison, both energy windows centered at 135 keV for 1^st and 2^nd acquisition provide a higher contrast than the others.

Close

Figure 3

Contrast of the six spheres with energy window

Export to PPT

Figures 4 and 5 show the effect of energy windows on the image quality of the simulated Jaszczak phantom with six hot spheres.

Close

Figure 4

Images of simulated Jaszczak phantom with a high-energy collimator using large energy windows (above) and narrow energy windows (below)

Export to PPT

Close

Figure 5

CNR of the six spheres

Export to PPT

We notice, as shown in Figure 6, that when the energy increases, the CNR decreases, and therefore, the image quality decreases. We observed the best CNR values for the first two energy windows in both acquisitions. In these windows, we can distinguish the two large spheres very well, but it is hard to distinguish the three smallest spheres. This seems to have been caused by increased background noise due to large penetration.

Close

Figure 6

CNR of the six spheres with energy windows

Export to PPT

Discussion

In Y-90 bremsstrahlung imaging, the image quality and quantification are limited due to the high levels of object scatter, collimator septal penetration, and collimator scatter. The parallel hole collimator and energy window optimization in Y-90 have been studied.[123456] In this study, we used Monte Carlo simulation SIMIND code to demonstrate how the image quality degrades as function of imaging parameters. We have evaluated the image quality considering the contrast and contrast to noise ratio (CNR). The simulation data indicates that the choice of the acquisition energy window for Y-90 imaging has a great effect on the image contrast and contrast to noise ratio. Figure 3 shows the high contrast values in 135 keV center for both acquisitions. We notice, as shown in Figure 6, that the best CNR values are obtained in the first three windows. The simulations also show that the penetration is a significant problem for HE collimator at HE. The HE collimator with an energy window between 120 and 150 keV was selected as optimal acquisition setting with consideration of the contrast and contrast to noise ratio (CNR) and also geometric, penetration, and scatter photons. The optimization of collimator and acquisition energy window leads to improve the quantitative accuracy and Y-90 bremsstrahlung SPECT image quality.

Conclusion

In this study, the obtained results showed that the HE parallel-hole collimator with energy window 120–150 keV conditions provides the best imaging performance based on contrast and CNR values. The optimization of these parameters leads to improved treatment efficacy and Y-90 bremsstrahlung SPECT imaging.