Inverse Methods for Planning Radiation Therapy

A new era in cancer radiation therapy has been gradually emerging during the last decade with the rapid development of many new and powerful methods for plan­ ning treatment and for optimizing the delivery of radiation doses. This process parallels in many respects the rapid devel­ opment in medical imaging technology that took place during the last 20 years, where computed tomography (CT), mag­ netic resonance imaging (MRI), and single photon and positron emission computed tomography (PET and SPECT) soon matured as universal diagnostic tools for clinical use. However, the parallels between the therapeutic and diagnostic developments are more profound than one might think at first sight. Both the diagnostic and the therapeu­ tic technologies allow a true three-dimen­ sional approach to be followed along the entire therapeutic chain, from diagnostic imaging to the delivery of the therapeutic effect by accurately shaped radiation beams incident on a tumour. But there are also deeper parallels between the two areas because the mathematical methods used in tomographic image reconstruction are similar to those adopted in some of the new methods for optimizing radiation therapy. This is seen most clearly in the analogy between the non-uniform dose delivery required by most of these meth­ ods and the back projection of filtered transmission or emission profiles used in image reconstruction algorithms [1, 2]. Owing to the existence of nuclide uptake, a distribution of photon attenua­ tion properties, or a proton density distri­ bution for SPECT, CT or MRI, respectively, these imaging techniques have the advan­ tage that there exists a true solution to the reconstruction problem, at least if all