COMPARISON OF DOSIMETRY METHODS TO OBTAIN RECTUM DOSAGE IN HDR-BRACHYTHERAPY OF CERVIX CARCINOMA

Background: Dosage measurements in brachytherapy of cervix carcinoma a re usually obtained with semiconductor dosimeters intrarectally and ca lculated using approximation methods for additional points, e.g. chose n according to ICRU Report 38. This procedure allows minimizing organ risk dosages and avoiding side effects. This study compares actual dos age measurements with computed approximations. Methods: In 75 applicat ions the measured actual rectum dosage was analyzed retrospectively. U sing graphic approximation methods in conjunction with the localizatio n radiographs the expected dose values at 5 detector points of the int rarectal semiconductor dosimeter and at the ICRU rectal reference poin t were determined. Prospectively for 11 additional applications the ex pected dosage :For various points within the rectum were computed duri ng therapy planning and additionally for specific reference points cor responding to Fletcher's lymphatic trapezoid and Chassagne's pelvic wa ll points. Results: The retrospective evaluation showed that 95% of va lues determined by graphic approximation methods varied by as much as +/- 30% from measured values. Factors causing errors were incorrect as sessment of the applicator's spatial positioning, non-orthogonal radio graphs, incorrect calibration of the semiconductor probe, movement of applicator and/or probe in the time between radiograph and application . In the prospective group 95% of deviations between measured and calc ulated values lay within an interval of +/- 40% (Figure 1). Possible s ources of error could be similar to those in using the graphic approxi mation, although the reconstruction of spatial positioning of the appl icator is possibly more exact. Doses determined at the ICRU rectal ref erence point were 5.6 +/- 2.5 Gy in the retrospective analysis and 6.1 +/- 1.6 Gy in the prospective study (Figure 2). The standard deviatio n of measured values at the specific reference points was +/- 30%. The mean dosage distribution was nearly symmetrical with regard to the bo dy axis, i.e. to the applicator position. Reasons for the relatively l arge standard deviation are e.g. difficulties in defining the referenc e points as well as in identifying them on the radiographs, also diffe rences in applicator positioning. Conclusions: The retrospective analy sis led to a larger error than the prospective one. The graphic approx imation method should only be utilized when computer-assisted treatmen t planning is not possible. Conspicuous are the differences between va lues obtained in computer planning and actual measured values. As thes e deviations cannot always be explained unequivocally, both computatio n and measurement should always be conducted in order to obtain an ade quate survey of dosage distribution within the rectum. Computer planni ng offers the additional advantage of determining the dose at various other reference points.