Evaluation of permanent I-125 prostate implants using radiography and magnetic resonance imaging

Abstract

The aim of this study is the evaluation of permanent I-125 prostate implants using radiography and magnetic resonance imaging (MRI). Twenty-one patients underwent radiography on the simulator and MRI within 3 days after implantation of the I-125 seeds. Isocentric radiographs were used for reconstruction of the seed distribution, after which registration with the seed-induced signal voids on MRI provided the seed positions in relation to the prostate. The prostate was contoured on the transversal magnetic resonance images, and dose-volume histograms were computed to evaluate the implants. The validity of the ellipsoidal prostate volume approximation, as applied in preimplant dose calculation, was assessed by comparison of ellipsoidal volumes given by prostate width, height, and length and prostate volumes obtained by a slice-by-slice contouring method, both on postimplant MRI. Prostate volume changes due to postimplant prostate swelling were assessed from radiographs taken at 3 days and 1 month after the implantation. The seeds were readily identified on T1-weighted spin-echo images and matched with the seed distribution reconstructed from the isocentric radiographs. The matching error, averaged over 21 patients, amounted to 1.8 +/- 0.4 mm (mean +/- standard deviation). The fractions of the prostate volumes receiving the prescribed matched peripheral dose (MPD) ranged from 32 to 71% (mean +/- standard deviation: 60 +/- 10%). Prostate volumes, obtained by the contouring method on postimplant MRI, were a factor 1.5 +/- 0.3 larger than the ellipsoidal volumes given by the prostate dimensions on postimplant MRI. Prostate volumes 3 days after the implantation were a factor 1.3 +/- 0.2 larger than the prostate volumes 1 month after the implantation. Registration of the reconstructed seed distribution and the MR images showed inaccuracies in seed placement, for example, two or more seeds clustering together or seeds outside the prostate. Registration of the reconstructed seed distribution and the MR images enabled evaluation of target coverage, which amounted to 60 +/- 10%. The discrepancy between prescribed dose and realized dose was caused by underestimation of the preimplant prostate volume due to the ellipsoidal approximation, postimplant prostate swelling at the time of evaluation, and inaccuracies in seed placement.