Development and validation of in vivo quantitative imaging methods for targeted radiotherapy treatment planning

Accurate in vivo dosimetry is essential for targeted radiotherapy (TRT). Organ doses are usually estimated using the MIRD-based schema which requires estimation of residence times (RT). These have typically been estimated using planar scans subsequently quantified using conventional planar (CPlanar) methods, which often do not provide the required accuracy. The goal of this research was to develop and validate new methodologies that reduce errors and variations in RT estimates. Since accuracy depends on several factors including imaging and processing methods, noise, organ volumes-of-interest (VOIs) definition, and anatomy and bio-kinetics variations over patients, we performed several experiments to study these. We first evaluated a quantitative SPECT (QSPECT) method. Since processing time for QSPECT is long, we developed a new quantitative planar (QPlanar) method which had accuracies approaching those of the QSPECT method. The results also demonstrated that precisions resulting from statistical noise were smaller than biases resulting from the imaging and processing methods. In order to study the effects of VOI definitions, we performed mis-registration and mis-definition experiments which demonstrated that the QPlanar method was no more sensitive to the mis-registration and less sensitive to the mis-definition of VOIs than the QSPECT method. Because estimation of RT builds upon estimation of activity at several time points, we evaluated three strategies: planar, SPECT, and hybrid methods combining planar with QSPECT. Since accuracy is affected by variations in bio-distribution and anatomy among patients, we generated a phantom population with realistic organ sizes and bio-kinetics. We showed that the QPlanar or hybrid QPlanar/QSPECT methods provided improved RT estimates compared to CPlanar methods, with accuracies approaching those of the QSPECT method. In summary, quantitative imaging methods were developed and evaluated. Among these methods, the QPlanar or hybrid QPlanar/QSPECT methods had accuracy and precision that approached those of the QSPECT method while providing simplified protocols and processing; these QPlanar methods thus may result in substantial improvements in dose estimation. This research also suggested that the order of relative importance of the various factors impacting accuracy and precision were: imaging and processing methods > patient variations > VOI mis-registration > VOI mis-definition ≈ noise.