Page 67 - ARNM-3-3

Advances in Radiotherapy
& Nuclear Medicine Software impact in Ho dosimetry
166

outside the liver. The characteristics of the patient sample Whole-liver and tumor absorbed doses revealed
are detailed in Table 1. statistically significant discrepancies between software
Hermia Voxel dosimetry showed mean liver and using the Wilcoxon-Test (p=0.002 for whole-liver dose and
tumor doses of 12 ± 4 Gy and 58 ± 23 Gy, respectively. In p<0.001 for tumor dose), with much lower doses in Q-suite
comparison, Q-Suite reported 44 ± 9 Gy for the liver and than using Hermia Voxel dosimetry. This discrepancy
209 ± 83 Gy for the tumor (Figure 3). suggested that something was not correct in one of the
methodologies associated with the software. Furthermore,
the Bland-Altman plot limits of agreement were notably
wide, indicating a high degree of variability and potential
discrepancies between the methods, as shown in Figure 3.
Estimated D values also revealed statistically significant
x
differences using both state-of-art software. The D , D ,
70
50
and D values were 55 ± 21 Gy, 41 ± 19 Gy, and 29 ± 16 Gy
85
for Voxel dosimetry, and 201 ± 78 Gy, 149 ± 67 Gy, and
108 ± 59 Gy for Q-Suite, respectively. The comparison is
represented in Table 2.
After a comprehensive analysis of all the steps involved
in generating the dose map using Hermes software, it
became evident that the significant underestimation of the
dose map was attributed to the relatively short time interval
considered in this study between the administration of
microspheres and the subsequent SPECT/CT acquisition.
Since the activity at the time of the imaging is above the
linear response of the gamma camera, the detectors will
underestimate the true count rate reaching it. This will
influence the perceived activity inside the patient and
Figure 3. Hermia Voxel Dosimetry and Q-Suite mean liver doses
(superior graph) and mean tumor doses (inferior graph) for each patient. therefore, the dosimetric calculations.
Patients No. 2 and 13 had two target lesions and, therefore, values for To validate this hypothesis, it was necessary to
each tumor are presented. Note that some of the mean tumor absorbed determine the optimal imaging time by calculating the
doses are below 150 Gy, the previously established minimum cut-off dose
required to advance to the therapy phase. This likely indicates some level maximum activity present at the moment of SPECT
of dose underestimation during the scout phase, influenced by various acquisition following the administration of 166 Ho
intra-procedure factors, such as consistent catheter placement. microspheres—ensuring the gamma camera response
remains within its linear range.
Table 1. Baseline characteristics of the patients
A 7 GBq vial of 166 Ho microspheres was positioned
Characteristics Value, n (%) or mean (range) on the SPECT’s bed (the acquisition time was 5 min) and
Sex static SPECT data was acquired over a range of activities,
Female 10 (77) at time intervals spanning several hours to a few days. For
each measured time point, the measured photon count rate
Male 3 (23) was plotted against the actual activity.
Age (years) 60 (39–74)
To characterize the relationship between the count
Number of lesions rate and activity in the system, the data were fitted using a
1 11 (84.6) paralyzable detector model, as described in :
9,21
2 2 (15.4)
Type of malignancy Table 2. DVH indices obtained with both software tools
Primary DVH indices Q‑suite (Gy) Hermia voxel dosimetry (Gy) p‑value
Hepatocellular carcinoma 10 (76.9)
D 50 201 (70–351) 55 (17–99) <0.0001
Secondary
D 70 149 (52–287) 41 (9–88) 0.0007
Neuroendocrine tumor metastasis 1 (7.7)
D 85 108 (40–250) 29 (5–75) 0.0007
Colorectal cancer metastasis 2 (15.4)
Abbreviation: DVH: Dose-volume histograms.

Volume 3 Issue 3 (2025) 59 doi: 10.36922/ARNM025220023