Reconstruction accuracy of a dedicated localiser for filmless planning in intra-operative brachytherapy

Background and purpose: With the use of HDR and PDR afterloaders containing a single stepping source, brachytherapy dose distributions can be optimised by varying the source dwell time. With the goal of implementing 'conformal brachytherapy', i.e. ensuring that the dose distribution conforms as accurately as possible to the target volume, we evaluated a set-up which enabled on-line implant localisation and dose planning during implantation. Materials and methods: The set-up, designated as an integrated brachytherapy unit (IBU), consists of a shielded operating room equipped with an HDR afterloader and a dedicated brachytherapy localiser connected to a treatment planning computer. The localiser is isocentric and has an extra degree of freedom in comparison to conventional simulators (i.e. an L-arm in combination with a C-arm) and enables viewing of the implant from any direction. A reconstruction algorithm which takes into account both rotation axes, i.e. the L-arm and C-arm angle, was developed for the localiser. The reconstruction procedure was tested by using the IBU localiser to measure the reconstruction accuracy with a phantom (containing 25 markers at well defined positions) and using reconstruction from radiographs. These results were compared to simulations where the accuracy of reconstruction was determined as a function of the reconstruction angle and the accuracy of read-outs of the localiser settings. On-line localisation and dose planning during implantation is based on filmless planning, i.e. fluoroscopy images and the corresponding localiser settings are imported into the treatment planning computer during implantation. The accuracy of filmless planning was determined using fluoroscopy images in the same set-up as for the experiments with the radiographs. The effect of reconstruction inaccuracies on the total irradiation time and the dose in target or normal tissue points was elucidated for clinically relevant implant geometries. The treatment plans of two phantoms based on reconstruction from films as well as fluoroscopy images were compared with plans for implants defined by exact co-ordinates. Results: The average reconstruction error due to the accuracy of the read-out of the localiser settings varied between -0.18 and 0.24 mm, with a standard deviation (arising from digitisation errors) ranging from 0.11 to 0.22 mm. Using filmless reconstruction and the 10 inch field of view of the image intensifier (without applying correction for the geometric distortions) the average reconstruction error ranged from 0.01 to 0.65 mm, and the standard deviation ranged from 0.40 to 0.73 mm. These errors arose as a consequence of the finite pixel size and geometric distortions. These limited errors did not influence the treatment time for clinical implant geometries and had only a minor effect (<1%) on the dose in markers during filmless planning. Conclusion: This IBU set-up, with a dedicated brachytherapy localiser, allows for a rapid and accurate filmless planning procedure based on implant localisation from fluoroscopy images.

• View at Publisher
• View at Pubmed
• View at Scopus
• View at Web of Science