Kidney dosimetry in [ 177 Lu]Lu-DOTA-TATE therapy based on multiple small VOIs

Purpose: The aim was to investigate the use of multiple small VOIs for kidney dosimetry in [ 177 Lu]Lu-DOTA-TATE therapy. Method: The study was based on patient and simulated SPECT images in anthropomorphic geometries. Images were reconstructed using two reconstruction programs (local LundaDose and commercial Hermia) using OS -EM with and without resolution recovery (RR). Five small VOIs were placed to determine the average activity concentration (AC) in each kidney. The study consisted of three steps: (i) determination of the number of iterations for AC convergence based on simulated images; (ii) determination of recovery -coefficients (RCs) for 2 mL VOIs using a separate set of simulated images; (iii) assessment of operator variability in AC estimates for simulated and patient images. Five operators placed the VOIs, using for guidance: a) SPECT/CT with RR, b) SPECT/CT without RR, and c) CT only. For simulated images, time -integrated ACs (TIACs) were evaluated. For patient images, estimated ACs were compared with results of a previous method based on whole -kidney VOIs. Results: Eight iterations and ten subsets were sufficient for both programs and reconstruction settings. Mean RCs (mean +/- SD) with RR were 1.03 +/- 0.02 (LundaDose) and 1.10 +/- 0.03 (Hermia), and without RR 0.91 +/- 0.03 (LundaDose) and 0.94 +/- 0.03 (Hermia). Most stable and accurate estimates of the AC were obtained using five 2mL VOIs guided by SPECT/CT with RR, applying them to images without RR, and including an explicit RC for recovery correction. Conclusion: The small VOI method based on five 2-mL VOIs was found efficient and sufficiently accurate for kidney dosimetry in [ 177 Lu]Lu-DOTA-TATE therapy.