A Voxel-based Curve Fitting Method for Reducing Imaging Time Points in Targeted Radionuclide Therapy Dosimetry

1436 Aim: Sequential images over multiple time points to obtain time-integrated activity (TIA) impose additional burdens on patients and clinical resources for targeted radionuclide therapy (TRT) dosimetry. This study aims to evaluate a curve fitting method for reduced imaging time points, i.e., single and two-time-point data, on voxel-based dosimetry based on In-111 Zevalin SPECT scans. Methods: A population of 9 XCAT phantoms with 3 anatomies and 3 In-111 Zevalin activity distributions were used in this study, modeling SPECT at scanning time point (TSC) of 1, 12, 24, 72 and 144 hr post tracer administration. An analytical projector of a medium energy general purpose collimator modeling attenuation, scatter, and geometric collimator-detector-response (GCDR) generated 128 realistic noisy SPECT projections, which were then reconstructed using the OS-EM algorithm (8 iterations and 16 subsets) with attenuation, effective source scatter estimation and GCDR compensation. The time-activity curves (TAC) for each voxel were fitted with a mono-exponential function in each phase using the nonlinear least squares method using 1- or 2- time point data, assuming zero activity at time 0 (MonoIter Method). After the last time point, physical half live (Tphy), 2 effective half-lives (Teff), including the mean Teff of the whole body (Teff_WB) and each specific organ (Teff_SO) over 9 phantoms were assumed for the TAC. For comparison, Madsen [1] and Hanscheid [2] methods were applied on the same data. In Madsen Method, mono-exponential equation approximation was based on mean Teff_WB and Teff_SO of the phantom population respectively. The area under TAC was then obtained to get the TIA for kidneys, spleen, liver, and lungs for all methods, which were compared with the standard, i.e., TIA obtained from sequential SPECT images over 5 time points based on the bi-exponential method, to get the absolute TIA error (TIAE), which were then compared using ANOVA test by SPSS for 1-time point study. Results: %TIAE is the smallest for 1 hr and 1 + 144 hr for 1- and 2-time point acquisition with our MonoIter Method, whereas %TIAE at 72 or 144 hr is much smaller for different organs as compared to other time points by Madsen and Hanscheid methods. For MonoIter Method, %TIAE is the smallest using Teff_SO, followed by Tphy and Teff_WB. %TIAE decreases substantially when data from another time point is added (<5.1% for all organs using Teff_SO). For the Madsen method, the %TIAE is smaller using Teff_SO as compared to Teff_WB. For the 1-time-point study, %TIAE is smallest by MonoIter Method for kidneys (6.30%) and spleen (12.96%), it is smallest for liver (6.97%) by Madsen Method, while it is smallest for lungs (11.53%) for Hanscheid Method. There is no statistically significant difference among different methods. Conclusion: For In-111 Zevalin SPECT, our proposed curve fitting method performed well for an early single time point or when the 2 imaging time points are well separated even without any prior Teff information. The %TIAE are <13% for all target organs, achieving similar accuracy as compared to 2 existing methods and with advantages for kidneys and spleen. Using two-time-point data substantially improved accuracy and may provide a better trade-off in terms of accuracy and clinical resource in TRT dosimetry. Further evaluations of our method using more clinical data and different applications are warranted. References: 1. Madsen, M.T., et al., Single time point dose estimate for exponential clearance. Medical physics, 2018. 45(5): p. 2318-2324. 2. Hanscheid, H., et al., Dose mapping after endoradiotherapy with 177Lu-DOTATATE/DOTATOC by a single measurement after 4 days. Journal of Nuclear Medicine, 2018. 59(1): p. 75-81. Research Support: Macau Science and Technology Development Fund (0091/2019/A2).