A Prospective Study Evaluating Registered Ultrasound and Fluoroscopy (RUF) for Intraoperative Dose Calculation: Improved Accuracy Compared to Current Ultrasound-based Intraoperative Dosimetry

Intraoperative transrectal ultrasound-based dosimetry during LDR prostate brachytherapy is imprecise due to sonographic distortion caused by seed echoes, needle tracks, and traumatic edema, all of which obscure seed positions or create false signals. We have previously described a system of ultrasound and fluoroscopy based seed localization (RUF) for intraoperative dose calculation, and sought to prospectively evaluate its accuracy in predicting post-implant dosimetry compared to standard ultrasound-based dosimetry. Fifty-two patients undergoing permanent Pd-103 seed implantation for prostate adenocarcinoma were prospectively enrolled between July 2011 and Sept 2013. Seed implantation was performed using standard ultrasound-based seed identification/localization method (USD) for intraoperative dose tracking. Intraoperatively upon implant completion, 3-6 fluoroscopic images were acquired using a non-isocentric mobile C-arm, as well as axial ultrasound images of the implanted prostate. These images were used for off-line verification of RUF dosimetry, and were not used for implant modification. Postimplant CT and MR scans were performed for each patient on post-op day 1 for contouring and dose calculation. Using the intraoperatively-acquired fluoroscopic images, registered ultrasound and fluoroscopy (RUF) was performed, wherein 3D seed cloud configuration is reconstructed, followed by registration of the seed cloud to ultrasound images using an image-based point-to-volume registration method that affinely registers the reconstructed seeds to the bright hyper-echoic regions of the TRUS volume. Prostate and urethral dosimetric parameters were separately calculated for analysis on RUF, USD, and CT/MRI image sets; rectal dose was not included due to impact of rectal probe on intraoperative compared to post-operative rectal volume. Squared differences between dosimetric measures for RUF vs. CT/MRI and USD vs. CT/MRI were calculated and root mean squared differences evaluated. Correlation coefficients and paired T-tests were used to independently evaluate pairwise similarity between USD, RUF, and CT/MRI data sets. Of the 52 patients enrolled, 47 patients had complete USD, RUF, as well as MR/CT image sets and were included in the analysis. RUF based real-time dosimetry showed significantly higher correlation with day 1 CT/MRI for all prostate and urethral dosimetric parameters compared to USD (Pearson correlation, P = 0.04). When compared to CT/MRI, RUF dosimetry showed no significant variation for all 7 dosimetric parameters analyzed, whereas US dosimetry varied significantly for 5 of 7 parameters (paired two-tailed T test, alpha < 0.01) [Figure 1]. Root mean squared differences from CT/MRI were smaller for RUF for all 7 parameters examined compared with USD. USD demonstrated a tendency to over-estimate dose to the prostate when compared to RUF. Intraoperative RUF approximated post-operative CT/MRI prostate and urethral dosimetry to a greater degree than the current ultrasound-based intraoperative method. RUF is deployable in combination with a standard non-isocentric C-arm, and demonstrates potential to minimize prostate underdosage not otherwise detected. A confirmatory phase II trial utilizing RUF for intraoperative iterative plan modification is underway.