Background: Conformal radiotherapy techniques as used in prostate treatment allow to spare normal tissue by conforming the radiation fields to the shape of the planning target volume (PTV). To be able to fully utilize the advantages of these techniques correct patient positioning is an important prerequisite. This study employing an electronic portal imaging device (EPID) investigated the positioning uncertainties that occur in the pelvic region for different patient positioning devices.
Patients and methods: 15 patients with prostate cancer were irradiated with or without rectal balloon/pelvic mask at a linear accelerator with multileaf collimator (MLC). For each patient multiple portal images were taken from different directions and compared to the digitally reconstructed radiographs (DRRs) of the treatment planning system and to simulation films (Table 1, Figure 1).
Results: In spite of different positioning devices, all patients showed comparable total positioning uncertainties of 4.0 mm (lateral), 4.5 mm (cranio-caudal) and 1.7 mm (dorso-ventral). The lateral positioning error was reduced for the pelvic mask patients while the cranio-caudal error increased (Table 2, Figure 2). A systematic and a random component sum up to the total positioning error, and a good estimate of the magnitudes of the two is possible from six to eight portal images (Figure 3).
Conclusions: With a small number of portal images it is possible to find out the systematic and random positioning error of a patient. Knowledge of the random error can be used to resize the treatment margin which is clinically relevant since this error differs greatly for different patients (Figure 4). Image analysis with EPID is convenient, yet has some problems. For example, one only gets indirect information on the movement of the ventral rectum wall (Figure 5). The successful operation of positioning devices, although, needs further improvement--especially if one focuses on IMRT.