Development of methods to assess the predictive power of pre-therapy imaging for selective internal radiation therapy

Abstract

Pre-therapy Technetium-99m (99mTc)-MAA SPECT may be used to predict the 90Y absorbed dose in selective internal radiation therapy (SIRT), while post-therapy 90Y bremsstrahlung SPECT may be used to measure the delivered absorbed dose. The primary aim of this work was to develop methods to overcome the differences in partial volume effects between pre- and post-therapy imaging and allow quantitative comparison of predicted and delivered 90Y absorbed dose. The secondary aim was to apply the methods to clinical data to assess the predictive power of 99mTc-MAA imaging. Measured 99mTc and 90Y PSFs were input into a PVC method and a novel resolution conversion method (RCM) to correct for differences in 99mTc and 90Y SPECT resolution. Comparisons were performed on absorbed dose maps derived from 99mTc and 90Y SPECT scans of phantoms and Monte Carlo simulated SPECT scans of patient data. The RCM and a bilateral filter were then applied to clinical data. The relationship between hepatotoxicity after SIRT and the predicted absorbed dose to the normal liver was investigated. It was shown that the 90Y PSF was best described by a function comprised of the sum of two Gaussian functions. The RCM and the bilateral filter were shown to elicit the most reliable results from the comparison of phantom and simulated reconstructed SPECT data. Analysis of patient data demonstrated that the mean absorbed dose to the normal liver could be predicted within 20%. However, the prediction of mean absorbed dose to tumours was unreliable, especially for small tumours. A relationship between the predicted absorbed dose to the normal liver and the liver toxicities was demonstrated. The methods developed allowed comparison of predicted and delivered absorbed dose distributions, derived from pre- and post-therapy SPECT images. The application to clinical data found that the absorbed dose to the normal liver could be predicted.