Evidence-based medicine has become the foundation in radiation oncology development. The generally accepted evidence is obtained from data accumulated from clinical practice. Outcomes data from various radiotherapy techniques are often analyzed with theoretical models. In a technology and science based medical discipline, significant variations exist in individual implementation of these technologies. Uncertainties may also arise from the limitations of the technology at various stages of development. The delivery of proton radiotherapy, for example, may pose uncertainties from additional sources and these uncertainties may be larger in quantity than those found in the delivery of photon radiotherapy, due to the fact that proton radiotherapy is still in the beginning phase of its technology development. Many mature technologies developed for photon radiotherapy, 3D IGRT, for example, are not yet widely implemented in proton delivery systems. These uncertainties may influence, to a critical extent, the outcome. There is, therefore, an urgent need to create a critical assessment of the uncertainties of technologies and their implementation in radiotherapy, and to develop methodologies to propagate these uncertainties so that they eventually relate to outcome.
This study explores methodologies of uncertainty propagation, including probability theory and Monte Carlo simulation. The uncertainty associated with the outcome models is evaluated and quantified with the measures of belief and plausibility from evidence theory. Furthermore, the Dempster rule of combination is used to fuse multiple models into a consensus model for better prediction. Clinical examples are used to illustrate the process.
Back to Mathematical Problems, Models and Methods in Biomedical Imaging