The impact of image-guided radiation therapy on the dose distribution in prostate cancer using deformable registration

Dosimetric uncertainties due to variable anatomy and beam setup variability pose a significant limitation in modern precision radiotherapy. These uncertainties may lead to discrepancies between the planned and actual dose distribution delivered to the patient. This may have an adverse impact on the treatment outcome in terms of recurrent tumour growth and/or causing complications in normal tissues.; This work investigates the hypothesis that image-guided radiation therapy is needed to reduce the detrimental effects of changes in anatomy on the delivered dose distribution in cancer patients. To test this hypothesis, a deformable model is developed to enable the quantification of dose differences due to patient repositioning and variable anatomy.; The deformable model is based on contour-driven thin-plate splines to track the position of tissue elements within the patient. This is combined with recalculation of the treatment plan using frequent computed tomography (CT) image data acquired at different times during treatment. It is demonstrated using a clinical prostate case that dose differences in the rectum and bladder are significant (∼25%) after a multiple fraction treatment.; The deformable model is validated using phantom and clinical prostate CT data. A mathematical phantom is used to demonstrate that the accuracy in tracking the dose delivered to a tissue element is 3--4% in high dose gradient regions. Ten prostate cancer patients with radio-opaque markers implanted in the prostate and seminal vesicles are used to demonstrate that the deformable model is accurate (∼2.5 mm) to within the intra-observer contouring variability.; The impact of correcting for setup uncertainty and inter-fraction tumour motion is explored by comparing treatment scenarios that would employ current image guidance technology to conventional treatment (i.e., alignment to external markers). This work demonstrates that geographic tumour miss is remedied using image-guided treatment and day to day variations in dose are reduced. However, normal tissue sparing may not improve unless the planning target volume (PTV) margin is reduced. With reduced margin, dose escalation based on multiple instances of anatomy through CT image guidance shows substantial discrepancies from using the static plan alone.; In conclusion, image guidance improves the precision of radiation therapy and the deformable anatomy modelling provides a powerful tool to define guidelines for when and how to implement adaptive radiation treatments.