Sci‐Fri PM: Delivery — 04: A Patient Scatter Model for In Vivo Radiation Therapy Verification Using EPID Dosimetry

Radiation therapy has become increasingly complex with the introduction of new technologies like intensity modulated radiation therapy (IMRT) and rotational‐IMRT. Thorough dosimetric verification is required to ensure sufficient tumour coverage and normal tissue sparing. Pretreatment verification is conventionally performed prior to a patient's course of treatment, but validation during treatment delivery does not usually occur. One method to determine whether the treatment was delivered correctly is through the comparison of a measured portal image (taken with an a‐Si EPID) to a predicted portal dose image (modeling the same EPID). In this work, a patient scatter model was incorporated into an existing fluence model to account for the effect of a patient during an in vivo portal image measurement. The patient/phantom CT data set is converted to an equivalent homogeneous phantom (EHP). The modeled beam fluence is then ray‐traced through the EHP and onto the EPID, accounting for patient attenuation. Patient scatter fluence is calculated through the superposition of a library of pre‐calculated Monte Carlo scatter fluence kernels. The dose delivered to the EPID is determined with the convolution of a series of mono‐energetic dose kernels. The patient scatter model was tested with slab phantoms for a range of thicknesses, air gaps and field sizes, and was found to accurately predict images within 2% and 3 mm. A prostate in vivo IMRT field prediction was also carried out, with a comparison of the relative images resulting in a model accuracy of 3% and 3 mm.