Evaluation of using small volume of interest regions for clinical kidney dosimetry in 177Lu-DOTATATE treatments.

Abstract

Segmentation of the whole-kidney parenchyma (WKP) is considered the reference method for kidney dosimetry of radiopharmaceuticals, as it provides the average absorbed dose to the fully delineated WKP. However manual segmentation of the WKP is time consuming, and automated segmentation requires operator verification and potential manual adjustments to the VOI. The aim is to determine if a small volume of interest (SV) method can generate similar kidney absorbed doses as the WKP method.

Methods: We obtained SPECT/CT of 18 patients at 24, 48, and 168 h after injection of [¹⁷⁷Lu]Lu-DOTATATE (7.3-7.8 GBq). The SPECTs were corrected for attenuation, scatter, and collimator detector response with Monte Carlo-based OSEM reconstruction (ASCC-SPECT) and post-filtered with a 0- to 12-mm Gaussian filter or were only attenuation corrected with a Hann post-filter (AC-SPECT). Kidney dosimetry based on the manually segmented WKP was used as reference method. Recovery coefficients (RCs) for each WKP were determined by Monte Carlo simulations, and normalisation factors, NFs, for SVs were determined relative to the WKP method. Kidney absorbed doses were estimated based on measured activity concentrations fitted using the mono-exponential function. The accuracy of the absorbed dose estimates for the SV methods, corrected with the NFs, were assessed using the standard deviation of the percentage difference in agreement with the reference method across all kidneys. Accuracy for kidney dosimetry using the SV method was calculated based on 1-5 VOIs with volumes of 4 mL (SV₄), 2 mL (SV₂), and 0.6 mL (SV_0.6).

Results: The mean RCs of the WKP volumes (31-243 mL) in non-filtered ASCC-SPECT and AC-SPECT were 0.85 (0.73-0.90) and 0.62 (0.46-0.51), respectively. In non-filtered images, the absorbed dose was overestimated by a factor of 1.22. However, applying a Gaussian filter with a kernel size of approximately 5 mm yielded absorbed dose estimates comparable to the reference WKP method. The accuracy of kidney dosimetry calculation based on one SV₄ on each SPECT data-point was 12%. The accuracy improved as the number of VOIs increased from 1 to 5. With the SV₂ method, using a mean of 5 VOIs per kidney parenchyma, the accuracy was 8.3%.

Conclusion: The small volume of interest (SV) method can provide absorbed dose estimates comparable to the whole-kidney parenchyma (WKP) method when optimized. Non-filtered images overestimated doses by 1.22, but applying a 5 mm Gaussian filter aligned SV results with the WKP method. Using multiple VOIs improved accuracy, with five 2 mL SVs achieving 8.3%. The SV method provides a less time-consuming alternative to WKP; however, its implementation is recommended to be validated and adjusted against a reference method.