[Poster] Influence of Patient Setup in the context of EPID-based In Vivo Dosimetry

F.Olivia¹, S. Celi^1,2, F. Husson PhD¹, J-C. Diaz¹
1 DOSIsoft, Cachan, France
2 Institut Curie, Medical Physics department, Paris, France

Presented at Denmark Congress 2014

Introduction: At present, the available commercial solutions for in vivo transit dosimetry reconstruct point dose values in the patient by applying back projection algorithms on portal images acquired during the whole treatment fraction. These solutions provide the means to compare the predicted dose to the measured dose at each point inside the patient, assuming the beams are aligned with the patient anatomy as planned. Therefore if the actual setup is not tested, reconstructed dose values may be erroneous in case of patient setup errors or breast deformation during treatment, even when the actually delivered dose is correct. The aim of this work is to analyze the possibility of using treatment portal images to assess a posteriori the actual patient setup to validate the reconstructed dose values in the patient or to adjust measured dose deviations by correcting the beam-patient alignment.

Conclusion: Image registration analysis and impact on EPID-based in vivo dosimetry. The “before/after” gamma-index test allows to identify the type of deformation in case of patient misalignment:

• If the GAI improves significantly after registration, the patient misalignment results from a major translation issue. Considering the computed translation, the early in vivo dose deviations have to be re-evaluated by reconstructing a more realistic back projection from the portal image.
• Else, a rigid transformation may not be the only reason why the points would not coincide. In vivo dose reconstructions must be interpreted with caution. The origin of the treatment
  reproducibility issue must be investigated – if large anatomical deformations are highlighted, the initial plan may be no longer valid.

With a suitable and automatic geometric processing, it is possible to improve the confidence in in vivo dosimetry results for tangential breast beams by reducing the effect of setup error during the course of treatment.