Optimisation of beam-orientations in conformal radiotherapy treatment planning

Many synergistic advances have led to the beginnings of the routine use of conformal radiotherapy. These include advances in diagnostic imaging, in 3D treatment planning, in the technology for complex treatment delivery and in computer assessment of rival treatment plans. A conformal radiotherapy treatment plan more closely conforms the high-dose volume to the target volume, reducing the dose to normal healthy tissue. Traditionally, human planners have devised the treatment parameters used in radiotherapy treatment plans via a manually iterative process. Computer 'optimisation' algorithms have been shown to improve treatment plans as they can explore much more of the search space in a relatively short time. This thesis examines beam-orientation computer 'optimisation' in radiotherapy treatment planning and several new techniques were developed. Using these techniques a comparison was performed between treatment plans with 'standard', fixed beam-orientations and treatment plans with 'optimised' beam-orientations for patients with cancer of the prostate, oesophagus and brain. Plans were compared on the basis of dose-distributions and in some cases biological models for the probability of damage to the target volume and the major organs-at-risk (OARs) in each patient group. A cohort of patients was considered in each group to avoid bias from a specific patient geometry. In the case of the patient cohort with cancer of the prostate, a coplanar beam-orientation 'optimisation' scheme led to an average increase in the TCP of (5.7±1.4)% compared to the standard plans after the dose to the isocentre had been scaled to produce a rectal NTCP of 1%. For the patient cohort with cancer of the oesophagus, the beam-orientation 'optimisation' scheme reduced the average lung NTCP by (0.7±0.2)% at the expense of a modest increase in the average spinal cord NTCP of (0.1±0.2)%. A non-coplanar beam-orientation 'optimisation' scheme was tested using five patients with tumours of the brain. The scheme reduced the mean and maximum orbit doses by (19±4)% and (4±3)%, when averaged over the patient cohort. The optic-nerve mean and maximum doses were reduced by (11±5)% and (12±1)% respectively. It was concluded that the 'optimisation' of coplanar beam-orientations led to improved treatment plans, but the improvements were relatively modest compared to those from the 'optimisation' of non-coplanar beam-orientations. (author)