177Lutetium SPECT/CT: Evaluation of collimator, photopeak and scatter correction

Abstract Purpose The goal of this study was to find the optimal combination of collimator, photopeak and scatter correction for 177Lutetium (177Lu) SPECT/CT imaging. Methods Three experiments [sphere‐to‐background ratios (SBR) 50:1, 10:1, and 2:1] were performed with the NEMA Image Quality phantom filled with 177Lu‐trichloride. SPECT/CT acquisitions were performed with the medium‐energy low‐penetration (MELP) collimator and 99mTc/Krypton collimator. For each acquisition six reconstructions, all with attenuation correction (AC), were made: the 113‐keV photopeak only, the 208‐keV photopeak only and both photopeaks combined, each with or without scatter correction (SC). Image quality was assessed using contrast‐to‐noise ratios (CNR), quantification accuracy by means of recovery coefficients (RCs) and the spatial resolution using line profiles. Results With SBR 50:1 and 10:1, both collimators met the Rose criterion (CNR > 5), whereas the MELP collimator showed a higher CNR for the 2:1 ratio. The RCmean was higher with the MELP collimator, most explicit after the 208‐keV AC/SC reconstruction for all acquisitions. The line profiles showed a better spatial resolution for the MELP collimator and the 208‐keV AC/SC reconstructions. Conclusion 177Lu SPECT/CT image quality and quantification was most optimal when acquired with the MELP collimator and reconstructed using the 208‐keV photopeak, with AC and SC.

metastatic prostate cancer is likely to further increase RNT. 2 In this respect, most centers use a "one-size-fits-all" approach for the administered amount of radioactivity during RNT, although treatment response between patients varies and patients might be undertreated due to this approach. 3 Dosimetry, which refers to the assessment of the absorbed dose in tissues, could aid in personalized RNT by increasing the dose to the tumor while minimizing irradiation of organs at risk. 4 Quantitative SPECT/CT imaging is important for 177 Lu dosimetry and although numerous articles are already available, a clear description for clinical practice is lacking and many sites use different methods. [5][6][7][8][9][10] Acquisition and reconstruction protocols are roughly based on collimator, photopeak definition, and corrections. The medium-energy low-penetration (MELP) collimator is advised for 177 Lu gamma imaging because of its lower septal penetration. 11 In our institute, a particular low-energy high-resolution collimator with thick septa ( 99m Tc/Krypton) is available, with similar specifications compared to the MELP collimator (Table 1). 177 Lu has two main photopeaks at 113 keV (6.2%) and 208 keV (10.4%). 7 According to the MIRD/ EANM guidelines, the 208-keV photopeak is preferable for imaging with MELP collimators and the 113-keV photopeak for LEHR collimators. 7 Combining the counts from both photopeaks boosts the overall signal, which could be beneficial in late imaging time points. 12 Next to the routinely used attenuation correction (AC) for SPECT/CT imaging, also scatter correction (SC) is suggested to improve the 177 Lu quantification. 13 As the 177 Lu photopeaks have quite different energies, multiple scatter windows are applied to correct for this image degrading effect.
Aside from absolute quantification of uptake, visual assessment of accumulation in tissues and delineation of lesions is important in clinical practice. Physicians are used to visually assess normal tissue to tumor ratios, where uptake in small tumors is hampered by image contrast, noise and spatial resolution. So, it is important to recognize that a compromise between these factors has to be made when selecting the optimal imaging and reconstruction parameters for 177 Lu dosimetry.
The goal of this study was to find the optimal combination of collimator, photopeak, and scatter correction for 177 Lu SPECT/CT imaging. Different protocols were compared with respect to contrast-to-noise ratios (CNR), quantification accuracy by means of recovery coefficients (RCs), and spatial resolution using line profiles.

2.A | Phantom image acquisition and reconstruction
with C the average number of counts in a certain volume with known activity concentration A. Image quality was assessed using CNR calculated according to Eq. (2): where C S represents the average number of counts in each sphere, C B the average number of counts in the background volumes used to determine the CF, and σ B the average standard deviation of the background volumes. 15 The Rose criterion, CNR > 5, was used to classify whether an object is detectable or not. 16 Absolute quantification was evaluated using the average recovery coefficient (RC mean ) according to Eq. (3)9:  Line profiles across the 37 and 17 mm spheres were drawn ( Fig. 4) and present comparable differences between both collimators compared to the RC mean -curves in terms of quantification. The line profiles derived from the MELP collimator are steeper, indicating a better spatial resolution for this collimator.

| DISCUSSION
The goal of this study was to find the optimal combination of collimator (MELP vs 99m Tc-Krypton), photopeak selection (113 vs 208 keV vs 113 + 208 keV), and scatter correction for 177 Lu SPECT/ CT image quality, quantification, and spatial resolution. Figure 2 shows that image quality using both the 113 + 208 keV photopeaks,  (Table 2). Furthermore, Fig. 2 shows that the MELP collimator meets this Rose criterion in smaller sphere sizes compared to the 99m Tc/Krypton collimator in SBR 2:1.
The highest RC mean in this series was 0.9 for the MELP SBR 50:1 and 208-keV AC/SC reconstruction in the 37-mm sphere (Fig. 3).
Overall, the recovery of the 208-keV photopeak only was the high- Contrast-to-noise over all spheres and reconstructions based on the average sphere and background activity concentration. The dotted line represents contrast-to-noise ratios (CNR) = 5 according to the Rose criterion. 14 The highest CNR can be observed from the medium energy low penetration collimator and the 113 + 208 keV attenuation correction/scatter correction reconstruction.
T A B L E 2 Contrast-to-noise ratios of the 37 mm sphere. | 275 compared to the 99m Tc/Krypton collimator. This is probably due to the difference in septal thickness of 1.14 and 0.4 mm, respectively (Table 1). An increased number of redundant high energy photons are passing the septa of the 99m Tc/Krypton collimator, resulting in images with high noise levels and a lower CNR (Fig. 2).
A limitation of this study is that only spheres with homogenous radioactivity distributions were used for analysis, which is less representative of a clinical distribution within a tumor. Evaluation of other target geometries and heterogeneous activity concentrations would be interesting before selecting a protoco; however, this is not common practice. Such an approach is also not essential in the comparison between collimators, photopeak windows and scatter correction. In this study also, the number of iterations and subsets in the reconstruction protocol was not varied, which might have had some influence on the study outcomes. Yet, the effects of these reconstruction setting are far less than the choices for collimators, photopeak windows and scatter correction.
To conclude, based on the reported results in this study, clinical 177 Lu SPECT/CT acquisitions in our institute are performed with the MELP collimator and the photopeak window is set at 208 keV with