Examination of dosimetry accuracy as a function of seed detection rate in permanent prostate brachytherapy

The variation of permanent prostate brachytherapy dosimetry as a function of seed detection rates was investigated for 1125 implants with seed activities commonly employed in contemporary practice. Post-implant imaging and radiation dosimetry data from nine patients who underwent PPB served as the basis of this simulation study. One-thousand random configurations of detected seeds were generated for each patient dataset using various seed detection levels from 30% to 99%. Dose parameters, including D90, were computed for each configuration and compared with the actual dosimetry data. A total of 108 000 complete sets of post-PPB dose volume statistics were computed. The results demonstrated that although the average D90 differed from the true value by less than 5% when 70% or more seeds were identified, the D90 of an individual case could deviate up to 13%. The 95% confidence interval (CI) of estimated D90 values differ by less than 5% from the actual value when 95% or more seeds are detected, or approximately a 7 Gy difference in the D90 value for a prescription dose of 144 Gy. Estimated target volume dose parameters tended to decrease with reduced seed detection rates. The most variable dose parameter was the prostate V100 in absolute scale while the urethral V100 was most variable in a relative sense. Based on this comprehensive simulation study, it is suggested that 95% or more seeds need to be localized in order to provide an accurate estimation of dose parameters for contemporary iodine 125 permanent prostate brachytherapy.