Last Updated: 14th August 2019

Verification of the Accuracy of Image-Based Dosimetry Methods used in SIRT Microsphere Therapy

Aim: To verify the accuracy of image-based dosimetry when used in 90Y-microsphere selective internal radiotherapy (SIRT) treatments. In addition, this study investigates the level of agreement between the distribution of MAA in pre-therapy investigations and the post-therapy 90Y-PET distribution, to see whether MAA is a good analogue for 90Y-microspheres in patients treated at King’s College Hospital (KCH).    Methods: Several software tools have been developed using MATLAB (The MathWorks, Inc., www.mathworks.com, version R2017a) to implement dosimetric calculations based on local energy deposition. Results calculated for three test patients have been compared to output from commercial software being used by KCH for SIRT dosimetry. Comparisons have been made between 90Y-microsphere activity measurements made using a PET scanner and using a radionuclide calibrator. A short literature review has been conducted to investigate the possibility of using in-vitro dosimetric techniques to help verify the accuracy of image-based dosimetry methods. Finally, an analysis of the similarity of MAA and 90Y distributions has been conducted, using a novel comparison method developed for this study.

Results: Dose calculations performed using the in-house developed software tools were generally found to be in excellent agreement with the commercial software tools used by the department. Discrepancies in the dose calculations have been explained by differences in the methods used to determine tissue density. Measurements of 90Y-microsphere sample activities using both a radionuclide calibrator and a PET scanner demonstrated significant differences, with PET-derived activities being 10-30% higher or 5-25% lower than the calibrator-derived measurement for resin and glass microspheres respectively. Poor agreement was seen between MAA-SPECT and 90Y-microsphere distributions, with differences in the level of agreement being noted for different disease types and different pharmaceuticals.

Conclusions: Whilst image-based dosimetry is straightforward to implement following SIRT treatments, the uncertainty associated with the calculated results is high, on the order of 20-30%. Much of this uncertainty is systematic, and accuracy could be improved with the introduction of traceable calibration sources for 90Y-microspheres.