Comparison of CT and MR-CT fusion for prostate post-implant dosimetry.

PURPOSE The use of T2 MR for postimplant dosimetry (PID) after prostate brachytherapy allows more anatomically accurate and precise contouring but does not readily permit seed identification. We developed a reproducible technique for performing MR-CT fusion and compared the resulting dosimetry to standard CT-based PID. METHODS AND MATERIALS CT and T1-weighted MR images for 45 patients were fused and aligned based on seed distribution. The T2-weighted MR image was then fused to the aligned T1. Reproducibility of the fusion technique was tested by inter- and intraobserver variability for 13 patients. Dosimetry was computed for the prostate as a whole and for the prostate divided into anterior and posterior sectors of the base, mid-prostate, and apex. RESULTS Inter- and intraobserver variability for the fusion technique showed less than 1% variation in D90. MR-CT fusion D90 and CT D90 were nearly equivalent for the whole prostate, but differed depending on the identification of superior extent of the base (p = 0.007) and on MR/CT prostate volume ratio (p = 0.03). Sector analysis showed a decrease in MR-CT fusion D90 in the anterior base (ratio 0.93 ±0.25, p < 0.05) and an increase in MR-CT fusion D90 in the apex (p < 0.05). The volume of extraprostatic tissue encompassed by the V100 is greater on MR than CT. Factors associated with this difference are the MR/CT volume ratio (p < 0.001) and the difference in identification of the inferior extent of the apex (p = 0.03). CONCLUSIONS We developed a reproducible MR-CT fusion technique that allows MR-based dosimetry. Comparing the resulting postimplant dosimetry with standard CT dosimetry shows several differences, including adequacy of coverage of the base and conformity of the dosimetry around the apex. Given the advantage of MR-based tissue definition, further study of MR-based dosimetry is warranted.