Joint modeling of missing and mismeasured measurements for computing radiotherapy margins

PurposePelvic organs are naturally prone to positional and volumetric changes over time. These individual organ changes may result in variations in the clinical target volume (CTV) position and shape. Planning target volume (PTV) that encompasses the CTV with some margins to account for such uncertainties in patient positioning, organ motion, and beam geometry is universally accepted today in radiotherapy. Keeping this view in mind, the present study has been planned in order to determine the setup error (random and systematic) in patients of carcinoma cervix and rectum image-guided radiotherapy and to propose a population-based three-dimensional CTV to PTV margins in IGRT scenario.Materials and methodsThe mean margin displacements of 61 patients measured through different directions. The proposed method is used to estimate margins of displacement of specific CTV dosing. A further method is illustrated to simulate the dose received by the CTV. The margin distance compared with Van Herk model through consideration of systematic and random errors. Different site-wise displacements were measured and compared. The CTV performance dosimetry has been carried through proposed model and compared with the traditional margin for each of the datasets.ResultsThe PTV margins obtained on lateral, longitudinal and vertical sides on Cervix site are 0.91, 1.93 and 0.96 [Van Hark's formula] and 0.84, 1.79 and 0.89 through proposed formula. It can be noted that in both the cases, the respective margin obtained through proposed method was lesser in comparison to that of Van Hark's method.ConclusionsThe presence of heterogeneity in patient's random error is natural. This random error is highly influential on radiotherapy margin widths of the patients. The Bayesian statistics are useful to deal with such a problem. In this article, a simple and well-defined model is adopted to deal with the patient's heterogeneity problem. It has been observed that the model accounted the variability in the margin estimation. The variation in margin can be extended from one cancer site to another cancer site.