Accuracy of lutetium-177 SPECT activity quantification and patient-specific dosimetry: a Monte Carlo study
MetadataShow full item record
The goal of radiopharmaceutical therapy (RPT) is to deliver the maximum dose to cancerous tumours while sparing healthy tissue. Ideally, the radiopharmaceutical should accumulate in the tumorous tissue and the radiation entirely absorbed for tissue destruction; however, this is not the case. Dosimetry strives to balance the efficacy of delivering the maximum dose to the tumour cells with minimal toxicity to the healthy tissue. Patient-specific dosimetry offers the potential for RPT to reach its full potential as a powerful precision-based treatment. Efforts for patient-specific dosimetry remain a challenge due to the steps involved in the clinical dosimetry workflow. SPECT/CT imaging allows the estimation of the bio-kinetic distribution of the radiopharmaceuticals with good precision, which is required for accurate dosimetry. Different dosimetry software is available (commercial and non-commercial), but these should be benchmarked before being used. Optimising the imaging process for activity quantification and dosimetry in the NM discipline is essential, and Monte Carlo (MC) techniques have been used successfully in this endeavour. Furthermore, full MC dosimetry gained wide acceptance as the gold standard and the most accurate means for patient-specific dosimetry. MC simulations offer the advantage of having a gold standard against which the dosimetry can be benchmarked, and the dosimetry accuracy evaluated. Therefore this thesis aimed to assess the accuracy of 177Lu SPECT activity quantification and patient-specific dosimetry using MC simulations. The focus was on the bio-distribution of 177Lu- DOTATATE due to its clinical relevance in RPT of patients with metastasised neuroendocrine tumours. The kidneys are the dose-limiting organs for RPT with 177Lu-DOTATATE. Studies have shown kidney doses to vary significantly between patients. Given the relation between tumour absorbed dose and tumour reduction, for a complete efficacy evaluation, absorbed doses should be determined not only for the kidneys but, where possible, extended to the tumours. This study incorporated voxel-based phantoms generated from phantom and patient CT data to perform virtual image-based activity quantification and dosimetry using MC simulations. The voxel-based phantoms were modified to include spherical structures mimicking tumours. The first objective of this study was to validate a model of the Siemens Symbia T16 dual-head SPECT/CT gamma camera available in our clinic using the SIMIND MC program for 177Lu imaging. The validation was achieved by comparing experimental and simulated gamma camera performance planar and SPECT criteria tests. The results were in good agreement and provided adequate confidence that SIMIND could emulate the Symbia T16 successfully and be used for further investigations of 177Lu SPECT/CT image quantification. The second objective investigated the effect of sphere and cylinder calibration factor (CF) geometries and their corresponding recovery coefficients (RCs) on the quantification accuracy of 177Lu SPECT images using MC simulations. The investigations were performed using geometries of a cylindrical, an anthropomorphic torso, and patient-specific phantoms. The quantification accuracy was evaluated for tumours and the kidneys. The results demonstrated that 177Lu SPECT quantification accuracies compared favourably for sphere-based and cylinder-based CF and RC combinations when all SPECT corrections were applied. The absolute quantification accuracy of ≤ 3.5% compared well to literature findings and complied with the 5% requirements for accurate dosimetry. The third objective of the thesis aimed to compare the accuracy of the absorbed doses computed with the software LundADose and OLINDA/EXM 1.0 using three patient-specific voxel-based phantoms. The dosimetry accuracy was assessed by comparing the computed doses to the “true” activity images combined with full MC dosimetry to define the gold standard. The accuracy between LundADose (6.6%) and OLINDA/EXM 1.0 (8.1%) was comparable. The ≤ 10% dosimetry accuracy suggested that the software platforms approximated the true dose estimates and advocated for the dosimetry accuracy to be reliable.