Quantitative nuclear imaging for dosimetry in radioembolization

Intra-arterial microsphere radioembolization is an increasingly applied technique for treatment of unresectable liver tumors. During radioembolization, microspheres (diameter 30 μm) loaded with a high-energy beta-emitting radionuclide, such as yttrium-90 (Y-90) and holmium-166 (Ho-166), are instilled in the hepatic artery via a catheter. The majority of microspheres will accumulate in and around the liver tumors, because these are almost exclusively perfused by arterial blood, whereas the healthy liver tissue mainly depends on the portal vein for its blood supply. Prior to treatment, a scout dose of technetium-99m macro-aggregated albumin (Tc-99m-MAA) particles, which serve as microsphere-surrogates and aim to predict the in vivo distribution of the treatment dose of microspheres, is administered to the patient for treatment planning. Image-based assessment of both the scout dose and the treatment dose distribution (for therapy evaluation) can be performed by the nuclear imaging techniques Single Photon Emission Computed Tomography (SPECT) and Positron Emission Tomography (PET). In this thesis, quantitative nuclear imaging methods are developed, evaluated, and exploited to enable accurate assessment of the in vivo Y-90 and Ho-166 microsphere distribution for dosimetry in radioembolization. It was found with phantom experiments and Monte Carlo simulations that Y-90 and Ho-166 gamma-camera images are heavily affected by photon scatter, photon attenuation, and penetration of high-energy photons through the collimator septa. Extensive correction for all of these image degrading effects is required to enable reconstruction of quantitatively accurate Y-90 and Ho-166 SPECT images. However, hindered by the broad range and continuous nature of the photon energy spectra, only approximate correction models are included in current state-of-the-art clinical Y-90 and Ho-166 SPECT reconstruction algorithms. In contrast, model-based correction techniques for scatter, random, and attenuation effects are clinically available for Y-90 PET imaging. Consequently, state-of-the-art clinical Y-90 PET images were found to be superior to state-of-the-art clinical Bremsstrahlung SPECT images for assessment of the microsphere distribution after Y-90 radioembolization. Motivated by the cost-effectiveness and wide availability of SPECT cameras, a new Y-90 Bremsstrahlung SPECT reconstruction algorithm was developed, incorporating a fast Monte Carlo simulator for correction of the image degrading effects. The Monte Carlo-based reconstruction approach substantially improved the quantitative accuracy of Y-90 Bremsstrahlung SPECT images, which may be used as an alternative to Y-90 PET images for post-treatment dosimetry in Y-90 radioembolization. The Monte Carlo-based reconstruction approach also improved the quantitative accuracy of Ho-166 SPECT images, as was demonstrated in phantom experiments. Furthermore, it was shown in a clinical Ho-166 radioembolization study that the in vivo distribution of both a scout dose and a treatment dose of Ho-166 microspheres can be accurately quantified withSPECT. A scout dose of Ho-166 microspheres was found to be a significantly better predictor for the mean radiation absorbed dose to the lungs, which is an important contra-indication for radioembolization, than a scout dose of Tc-99m-MAA microsphere-surrogates. These results indicate that treatment planning in radioembolization is improved by using the same quantifiable microspheres for scout dose and treatment.