A Comparison of AUA Symptom Scores following Permanent Low-Dose-Rate Prostate Brachytherapy with Iodine-125 and Cesium-131

Purpose: Transrectal ultrasound (TRUS) is the most common image modality used in permanent prostate brachytherapy (PPI). Although providing excellent prostate visualization, TRUS images are inferior to MRI images for identifying critical structures such as neurovascular bundles (NVB). To preserve erectile function after PPI, the traumatic injury resulting from needles passing through NVB is worth investigating. Our goals were to develop a novel method to register post-implant MRI to pre-plan ultrasound (US) images and demonstrate its usage on evaluating needle passages to NVB. Materials and Methods: A series of transverse TRUS images of prostate and several non-isocentric C-arm fluoroscopy (FL) images are taken intraoperatively right after PPI. The registration of post US images and reconstructed 3D seed cloud from FL images (seeds_FL) will be used as a bridge to register post MRI to pre US images. The Registration of Ultrasound and Fluoroscopy (RUF) images is done by an intensity-based point to volume algorithm. A 1D scaling factor along anterior-posterior direction is used to account for prostate deformation with presence of ultrasound probe. The day one post CT and MRI images are co-registered and the 3D seed cloud is segmented from CT (seeds_CT) whereas critical structure contours obtained from MRI (T2). The iterative closest point (ICP) algorithm is used to compute rigid transformation between seeds_FL and seed_CT. All contours from MRI can be transferred to FL coordinate using registration of two seed clouds and then to US coordinate using RUF registration. Because the intraoperative pre-US images share the same coordinate with post US images, the post MRI is then registered to pre-US after this two step transformation. To demonstrate this registration method, post MRI from twelve patients (mean prostate volume 34 9 cc, only intraprostatic seeds implanted with parallel needles) were registered to pre-US. The prostate, urethra, rectum, penile bulb and NVB contours obtained from MRI were overlaid on US images. The overall registration accuracy was evaluated by comparing prostate, urethra and rectum contours from both data sets. The planning needle interferences with NVB were investigated. Results: After registration, in 8 of 12 patients, the prostate base, apex and center of mass of both prostate and urethra match very well (!0.3 cm) between two contour sets. The anterior surface of rectum matches well within these 8 patients too. Large rotational errors (O10o) presented for two cases due to failure of the ICP algorithm, and about 1 cm base to apex direction offset presented in another two cases mostly due to offset in RUF registration. The penile bulb contours from two data sets generally off 1 cm which shows our registration method only works with structures close to prostate since seed locations in prostate was part of registration algorithm. When planning needles were superimposed over contours, there were about 3-4 needles passing through or in close vicinity of each NVB in all patients. The needles and NVB interference mostly happens under apex part of prostate. Conclusions: A novel method to register post-implant MRI to pre-plan US images was developed and demonstrated in accessing needle passages to NVB during PPI. The pattern of needle and NVB interference obtained can be used to design generic non-parallel implant technique to avoid traumatic injury to NVB. The registration information from fusion of post-implant MRI and preplan US can also provide insights to develop robust registration method between pre-implant MRI and US without using seeds as the bridge.