Dosimetry for Y-90 Radioembolization using a Long axial Field of View PET/CT scanner: Impact of Image Reconstruction and Scan Duration on Determined Absorbed Dose and Compliance with Prescribed Dose

Ziel/Aim: The increased sensitivity of long axial field of view (LAFOV) PET/CT scanners compensates for the low branching ratio of Y-90, enabling accurate dosimetry. Here, we evaluate the impact of variations in the image reconstruction towards a shorter scan time on dose maps and assess the compliance of prescribed and absorbed doses.

Methodik/Methods: 19 patients (6 NET, 5 CRC, 2 HCC, 2 lung cancer, 2 breast cancer, 1 ICC, 1 CUP) receiving a total of 27 treatment cycles underwent 30 min scans on a Siemens Biograph Vision Quadra®​ LAFOV PET/CT 15 to 43 h post radioembolization. Image reconstruction was varied between iterations of 2i5s and 4i5s for 30 and 20 min (rebinned) listmode data. Dose maps generated via a Monte Carlo simulated dose kernel were analyzed for mean dose (D_mean) in tumor, parenchyma and total liver. Dose-volume histogram (DVH) metrics (D75, D50, V40) of one patient supported the evaluation. D_mean was assessed to determine compliance (± 10%) with prescribed doses.

Ergebnisse/Results: Increasing iterations from 2i5s to 4i5s resulted in a mean deviation of D_mean by 2.0±1.1% (tumor), 1.8±3.0% (parenchyma) and 0.8±1.3% (total liver). Maximum change in D_mean was 5.7% (tumor), -11.1% (parenchyma) and 4.2% (total liver). Reduction of scan time to 20 min resulted in a mean deviation of D_mean by 0.6±0.6% (tumor), 0.7±1.2% (parenchyma) and 0.7±1.1% (total liver). Maximum change in D_mean was -1.6% (tumor), 4.0% (parenchyma) and 4.2% (total liver). The exemplary patient’s tumor had a D50 of 144 Gy (2i5s) and 145 Gy (4i5s), a D75 of 110 Gy (2i5s) and 108 Gy (4i5s), and liver V40 was 51.8% (2i5s) and 50.6% (4i5s). Comparison with prescribed doses showed compliance for tumor and parenchyma in 31% of the patients. In 44%, D_mean in the tumor did not meet the prescribed dose (mean deviation -34.3±14.8%). In 33%, parenchymal dose limits were exceeded (mean deviation 46.4±35.5%; maximum deviation 63 Gy vs 30 Gy; all<80 Gy).

Schlussfolgerungen/Conclusions: The D_mean across tumor, parenchyma and total liver derived from dose kernel convolution were minimally affected by iterations and remained consistent down to 20 min. This dosimetric method enabled a preliminary assessment of prescribed versus absorbed dose, with ongoing detailed DVH analysis for a larger patient cohort. We plan to compare the accuracy of these dose maps with patient-specific Monte Carlo simulations.

Publikationsverlauf

Artikel online veröffentlicht:
12. März 2025