Multiple vertebrae improves precision in image-based bone marrow absorbed dose estimation in [177Lu]Lu–DOTATATE treatment

BACKGROUND: Lu]Lu-DOTATATE. Bone marrow absorbed dose estimates and their precision were analyzed based on the number of vertebrae included in the calculation. Additional evaluations were performed using both measured (Lung-Spine phantom with spherical inserts) and simulated (digital XCAT phantom) SPECT data, with noise levels matching patients' bone marrow. RESULTS: Increasing the number of vertebrae from one to six improved precision in absorbed doses, reducing the coefficient of variation (COV) from 34 to 6.2%. Only minor differences (0.1%) in bone marrow absorbed dose were observed when adjusting reconstruction parameters from 1 subset, 60 iterations, to 12 subsets, 5 iterations. In the Lung-Spine phantom, smaller volumes (4 mL) were more sensitive to noise than larger volumes (16 mL), with COVs of 36% and 17%, respectively, at 60 updates. In the XCAT phantom, a decrease in recovery was observed in vertebrae near high-uptake regions (L5: 0.63, L1: 0.29), at 60 updates at the highest noise level. CONCLUSION: Including multiple lesion-free vertebrae enhances precision in image-based bone marrow absorbed dose calculations. Nevertheless, careful selection of vertebrae is important, as closely located high-uptake areas can bias absorbed dose estimates. Applying partial volume correction could mitigate spill-out effects, further improving accuracy. Additionally, the minimal differences between reconstruction parameters suggest that bone marrow absorbed dose estimates remain stable across subset configurations.