The Robustness of Various Intelligent Models in Patient Positioning at External Beam Radiotherapy

Patient setup optimization has been required to fill the gap between individualtreatment and uncertainty in the external beam radiotherapy at each of thetreatment sessions. This uncertainty error consists of patient body misalignmentsand patient body displacement between different fractions. In this study, the patient geometrical set-up has been simulated comprehensivelyby 4D XCAT anthropomorphic phantom where the XCAT phantomwas used to access 4D modeling of dynamic organs motion. All of the possibleroto-translation displacement parameters that were effective on instigate patientposition before re-alignment were considered. While the data set was assembledfrom XCAT phantom including 2D translation and 2D rotation, the parallelismsof the dada set between position of the external markers and reference point (patientcouch) were considered. Moreover, the experimental validation models forfurther investigation were considered. For this aim, the captured data from XCATphantom was extended to four real patients. In some clinically available strategies,the corrective models have been implemented to estimate patient displacementof patient setup. In this study, four intelligent models were proposed for set-up,realignment, and continuous tracking of the patient positioning. Final results illustrate that Adaptive Neuro Fuzzy Interference Systemwith all markers can estimate the true patient position with less error. In this study, the four intelligent models were demonstrated to investigatethe robustness of various intelligent models in re-alignment and patientset-up at external beam radiotherapy. Finally, our correlation model “ANFIS” canestimate the true patient position with less error.