THE EVALUATION OF OPTIMIZED IMPLANTS FOR IDEALIZED IMPLANT GEOMETRIES

The purpose of this paper is to investigate the utility of implant qua lity measures on single stepping-source brachytherapy treatment plans. Four dwell weight optimization algorithms were applied to four regula r geometric implants: single plane, double plane, cuboid, and cylindri cal. The dwell weight optimization schemes included equal weighing, tw o commercial optimization schemes (dose-point and geometric) and a var iation of the Paterson-Parker distribution rules. The implant quality measures were investigated as a function of dose-per-integrated refere nce air kerma (IRAK) to eliminate bias resulting from a prescription c hoice. A particular dose per IRAK refers to a dose surface that is a f unction only of the relative dwell weight distribution and is therefor e well suited to investigate dwell weight optimization schemes. The im plant quality measures included the dose-nonuniformity ratio (DNR) dev eloped by Saw and a coverage index to assess the isodose coverage rela tive to the implanted volume. These were termed direct quantities due to their clear clinical significance. Additional measures include the ratio of the implant dose-volume histogram (DVH) to that of a point so urce exhibiting the same IRAK (R(p)) and the ratio of the optimized DV H to the equally weighted DVH (EWR). The widths of the R(p) curves and depths of the EWR curves were used to characterize these indirect imp lant quality measures. To evaluate the effectiveness of both the direc t and indirect measures, they were correlated with the DNR for an isod ose surface that covered the implant (D-0). The efficiency of the dwel l weight distribution was examined by noting the dose-per-IRAK surface D-0. The DNR exhibited a distribution with a minimum value in each im plant and optimization method. At this point the high-dose volume is m inimized relative to the prescription volume and choosing this dose as a prescription isodose will provide a relatively homogeneous dose dis tribution. However, the minimum DNR value did not provide a clinically useful implant coverage with most optimization schemes. The exception was the dose-point optimization that yielded an adequate coverage at the DNR minimum. The EWR curve exhibited a dip (at dose D-n) for most of the optimization schemes which was deepest for the dose-point optim ization. There was no direct correlation between the EWR(D-n) and homo geneity, but a large value of EWR(D-n) consistently predicted a poor h omogeneity. An examination of the coverage versus the DNR showed that in all cases, a tradeoff existed between coverage and dose homogeneity . In all. cases the dose-point optimization provided the best compromi se between coverage and homogeneity in addition to the most efficient implant. The application of these implant quality measures allowed an examination of the inherent quality of each dwell weight distribution, a task that would be very difficult without this type of guidance. Wh ile the indirect quality measures provided some properties that correl ated with the direct quality measures, further study is necessary befo re their role in dose distribution analysis is completely understood. Use of the dose-per-IRAK as the independent variable divorced the anal ysis from an arbitrary prescription criterion.