The effect of geometric errors in the reconstruction of iridium-192 seed implants

In the treatment of tumors using interstitial implants of radioactive seeds, the accuracy of computed dose distributions depends upon the accuracy with which the three-dimensional source geometries are reconstructed from radiographs of the implants. The effect of geometric reconstruction errors in iridium-192 seed implants were studied, using tumor dose as the measure. Tumor dose was defined as the average dose around the periphery of the treatment volume. Three ideal mathematical implants and five actual patient implants were used. The implants were distorted by randomly moving a specified number of seeds a specified distance. Tumor doses were directly calculated for the ideal implants. For the actual implants, isodose distributions were plotted and were read by a radiotherapist. For both types of implants, percentage errors in the tumor doses were calculated for the distorted reconstructions relative to the correct reconstructions. It was found that the tumor dose was accurate to within 5% if all the seeds were reconstructed to within 0.5 cm of their actual positions. Furthermore, up to 5% of the seeds could be mismatched between films, or otherwise incorrectly reconstructed, with position errors as large as 20 cm, and not change the tumor dose by more than 5%.