Purpose: Comparison of predicted portal dose images (PDIs) with PDIs measured with an electronic portal imaging device (EPID) may be used to detect errors in the dose delivery to patients. However, these comparisons cannot reveal errors in the MU calculation of a beam, since the calculated number of MU is used both for treatment (and thus affects the PDI measurement) and for PDI prediction. In this paper a method is presented that enables "in vivo" verification of the MU calculation of the treatment beams. The method is based on comparison of the intended on-axis patient dose at 5 cm depth for each treatment beam, D5, with D5 as derived from the portal dose Dp measured with an EPID. The developed method has been evaluated clinically for a group of 115 prostate cancer patients.
Methods and materials: The patient dose D5 was derived from the portal dose measured with a fluoroscopic EPID using (i) the predicted beam transmission (i.e., the ratio of the portal dose with and without the patient in the beam) calculated with the planning CT data of the patient, and (ii) an empirical relation between portal doses Dp and patient doses D5. For each beam separately, the derived patient dose D5 was compared with the intended dose as determined from the relative dose distribution as calculated by the treatment planning system and the prescribed isocenter dose (2 Gy). For interpretation of observed deviating patient doses D5, the corresponding on-axis measured portal doses Dp were also compared with predicted portal doses.
Results: For three beams, a total of 7828 images were analyzed. The mean difference between the predicted patient dose and the patient dose derived from the average measured portal dose was: 0.4+/-3.4% (1 SD) for the anterior-posterior (AP) beam and -1.5+/-2.4% (1 SD) for the lateral beams. For 7 patients the difference between the predicted portal dose and the average measured portal dose for the AP beam and the corresponding difference in patient dose were both greater than 5%. All these patients had relatively large gas pockets (3-3.5 cm in AP direction) in the rectum during acquisition of the planning CT, which were not present during (most) treatments.
Conclusions: An accurate method for verification of the MU calculation of an x-ray beam using EPID measurements has been developed. The method allows the discrimination of errors that are due to changes in patient anatomy related to appearance or disappearance of gas pockets in the rectum and errors due to a deviating cGy/MU-value.