ESTRO 38 Abstract book

S585 ESTRO 38

phantom on ten vaginal treatments previously delivered at our Institution. Results The characterization of the MVA system showed no significant dependence on radiation intensity for a range of Air Kerma Strengths between 34 and 13 mGym 2 h -1 . After its characterization and calibration, the MVA system demonstrated the ability to resolve the HDR BT source position and dwelling time along the first 70 mm of the applicator, encompassing the most clinically relevant region for HDR BT vaginal treatments. 95% of measurements showed positional discrepancies between nominal and measured values within 0.8 mm and 1.5 mm for the source dwelling in the central and in the peripheral catheters, respectively. 95% of the measured dwell times for the central and peripheral catheters were within 0.4 s and 0.2s, respectively, if compared to the nominal values. Nominal and measured values for the source dwelling in the central catheter are shown in the figure.

Purpose or Objective I-125 seed implants are a reference treatment for low risk prostate tumors. The quality of this treatment depends on the precision in the placement of these seeds in the planned points to achieve the proper dose distribution. Post-implant dosimetry is based on the need to evaluate the quality of the implant once the forces applied by needles and by the ultrasound probe have disappeared and once the edema caused in the implant process has gone. In many places post-implant dosimetry is not done because it is time-consuming and also because of the difficulty to have reliable image modality to simultaneously define the position of the seeds and volumes: prostate and OARs. We developed a fast and reliable method to identify strands and individual seeds that would give geometric quality index of the implant, which allows us to check the robustness of the implantation method and if any cold spot is to be expected. Material and Methods The treatment planning system PSID™ 5.0 (Eckert & Ziegler BEBIG GmbH) is used, that allows us to extract the positions of the seeds of the planning in real time (online planning), as well as the positions after a month (postplaning). The MATLAB Statistics and Machine Learning Toolbox (MATLAB R2018a, MathWorks®) was used to write a code for seeds and strands identification using clustering techniques following the next steps: 1. Rigid registration with Iterative Closest Point, which uses an affine transformation to get rid of the differences between imaging datasets (online and postplan) due to its different reference systems. Hierarchical clustering with the dataset online using the MATLAB linkage function. Assignment using KDTreeSearcher and knnsearch MATLAB functions, first to generate a binary tree and then the subsequent k nearest neighbor search. Results The assignment is correct in most cases and it considerably reduces the time to obtain a visualization of the geometric differences between online and postplan datasets. Nevertheless the assignment is not completely when: there are seeds too far away from their original position in the XY plane, and when some strands appear bended in the Z direction. Once the assignment problem was solved we calculated the displacement vector for each seed, the strand centroid displacement vector, and the separation between consecutive seeds in the same strand to see if the 1 cm metric remains, which is the theoretical separation between seeds. We saw that in average the postplan consecutive seed separation was 0.9 cm. 2. 3.

Conclusion The proposed intracavitary MVA system has demonstrated the ability to simultaneously deliver and verify the HDR BT treatment with, in general, sub-mm and sub-second accuracy. Once improved, the system will provide essential information in real time to help ensure that the treatment is delivered in accordance with the planned treatment plan. Further studies may help to further increase spatial and time resolution. PO-1051 quality analysis automation of a prostate low dose rate seed implant using clustering methods D. Rodríguez Latorre 1 1 Hospital Universitario Central de Asturias, Radiofísica, Oviedo, Spain