Evaluation of Prostate Volume and Shape Change after Permanent Prostate Brachytherapy Using a Novel Deformable Registration Method

Prostate volume and shape change both during and after permanent prostate brachytherapy (PPB), mainly due to edema. Many studies have investigated how prostate volume changes from day 1 to day 30 or later. There is no available information on how prostate volume changes from immediately post-implantation to post-implant imaging. When determining implantation quality at end of procedure using intraoperative dynamic dosimetry techniques, the knowledge of potential prostate volume and shape changes from end of implantation to time of post-implant imaging would be valuable. In this study, we investigated prostate volume and shape changes after PPB using a novel deformable registration method based on implanted seeds. In 53 patients undergoing PPB, 3-6 non-isocentric C-arm fluoroscopy (FL) images were taken intraoperatively immediately after seed implantation. The 3D seed cloud from FL images (seeds_FL) was reconstructed. The post-implant CT and MRI images acquired on day 1 were co-registered with each other. The 3D seed cloud was segmented on CT (seeds_CT), whereas prostate contours were obtained from MRI (T2). The iterative closest point (ICP) algorithm is used to compute rigid transformation between two seed clouds (seeds_FL and seed_CT) and points-to-points corresponding. Prostate contours from MRI can be transferred to day 0 intraoperative coordinates using the registration of the two seed clouds. Due to prostate edema and different imaging position, prostate and adjacent anatomy deform between intra-operative day 0 and day 1 imaging. Point to point transformation from two seed clouds were used to warp post implant MRI and prostate contours through a Thin-Plate-Spline algorithm (TPS). Using the implanted seeds as a bridge, day 0 prostate contours can be obtained from reliable day 1 post MRI contours using above described deformable registration method. Intraoperative ultrasound images were not used to measure prostate volume due to artifacts from implanted seeds. The prostate volume and sizes of each axis were compared between day 1 MRI contours and day 0 wrapped contours. Twenty (38%) patients show average of 6%±5% (2.2cc±1.7cc) prostate volume increase between day 0 and day 1, while 33 (62%) patients show 5%±4% (2.5cc±2.4cc) prostate volume decrease. The average day 0 prostate volume is 34.3 cc ± 12.0 cc for patients with volume increase, smaller than 42.9 cc ± 11.0 cc for patients with volume decrease (p=0.005). For patients with volume increase, there is significant expansion only in AP direction (AP: 4mm±3mm) with p<0.001. For patients with volume decrease, there is shrinkage in all three dimensions (LR: 2mm±2mm, AP: 5mm±4mm, SI: 2mm±3mm) with p<0.001. Prostate volume and shape change between day 0 end of implantation and day 1 post imaging was evaluated using a novel deformable registration method. The prostate volume can increase or decrease between end of implantation and day 1 CT/MRI, indicating edema either reaches its maximum intraoperatively, or still continues development after implantation. The ability to characterize immediate postoperative prostate volume and shape changes allows for correlation of intraoperative dynamic dosimetry to post-implant CT dosimetry. Future work will analyze the effect of these variations on the degree of correlation between intraoperative and post-implant dosimetric outcomes.