Room: Karl Dean Ballroom B1
Purpose: To investigate the feasibility of a novel MRI-only workflow for post-implant dosimetry of prostate low dose rate (LDR) brachytherapy on prostate phantoms and patient.
Methods: Four realistic prostate phantoms (CIRS Inc.053L) were implanted with 321 stranded dummy I-125 seeds (IsoAid Advantageáµ€á´¹) according to realistic plans generated in VariSeed. The patient had 78 implanted seeds. The scans were performed on 1.5T (Philips Ingenia) MR scanner using a 3D multi-echo gradient echo sequence with in-phase echo times. To generate positive contrast for seeds, a novel Quantitative Susceptibility Mapping (QSM) was proposed. Seed positions were identified using unsupervised machine learning algorithms based on density-based spatial clustering of applications with noise (DBSCAN) and k-medoid clustering. The estimated seed positions and resulting DVH curves were compared to the clinical CT-based method in MIM software (MIM Symphony Dxáµ€á´¹).
Results: In phantoms, the seed shape was nicely reconstructed and all 321 seeds were correctly detected; the average distance between CT and MR-derived centroids was 0.4Â±0.3mm. No significant difference was found between the DVH curves (P<0.0001). Based on the same contours (on CT), the average difference between D90,V100,D2cc(rectum) and D2cc(urethra) was less than 0.2%. In patient although the physiological noise and lower resolution degraded the quality of the reconstructed susceptibility map, 77 seeds were correctly localized and only one seed was misdetected. The average distance between CT and MR-based seed positions was 2Â±2.6mm; no significant difference was observed between the DVH curves (P<0.005). The proposed workflow overestimated V100 by 0.7% and underestimated the D90 and D2cc (rectum) by 0.5% and 1.9% respectively.
Conclusion: This work indicates the great potential for replacing the conventional CT-based post-implant dosimetry which suffers from suboptimal soft-tissue visualization with the proposed MR-based workflow with superior soft-tissue contrast. Further optimization of the algorithm is required to improve the efficiency of the workflow in patients.
Not Applicable / None Entered.