ESTRO 2024 - Abstract Book

S3952

Physics - Image acquisition and processing

ESTRO 2024

Natalia Tejedor 1 , Núria Jornet 1 , Pedro Gallego 1 , Gemma Sancho 2 , Ana María Soto-Cambres 2 , Arantxa Mera 2 , Jaime Pérez-Alija 1 , Cristina Ansón 1 , Helena Vivancos 1 , Agustín Ruiz 1 , Marta Barceló 1 , Fátima Leo 1 , Alejandro Domínguez 1 , Víctor Riu 1 , Pablo Carrasco 1 1 Hospital de la Santa Creu i Sant Pau, Medical Physics, Barcelona, Spain. 2 Hospital de la Santa Creu i Sant Pau, Radiation Oncology, Barcelona, Spain

Purpose/Objective:

In recent years the use of magnetic resonance (MR) imaging as the only imaging modality for radiotherapy treatment planning (MR-only workflow) has expanded rapidly due to the development of new commercial software. Recently, our center has incorporated a specific platform for MR-only simulation and planning in prostate cancer. The aim of this study is to assess the dosimetric accuracy of the synthetic CT (s-CT) generated by the software.

Material/Methods:

Nine patients were included in the study. Patients were imaged on a Philips Big Bore CT scanner and subsequently on a Philips 3T Ingenia Elition MR scanner using the Ingenia MR-RT platform. MRCAT Pelvis algorithm was applied to generate a s-CT for dose calculation. Two independent set of structures were delineated in the s-CT and p-CT by a radiation oncologist. VMAT treatment plans were optimized and calculated on s-CT using Eclipse version 15.6.05 TPS with AcurosXB Dose to Medium algorithm. To minimize geometric differences between both imaging modalities, the planning CT (p-CT) was resampled to match s-CT voxel size. Deformable image registration between p-CT and s-CT was also performed by using SlicerRT Elastix algorithm. The treatment plan was recalculated on the deformed p-CT (p-CT def ). Dosimetric comparison of both plans was performed evaluating clinically relevant DVH points such as D mean , D 98% and D 2% for CTV and PTV and D mean and V 58Gy for organs at risk (OARs) such as bladder, rectum and femoral heads. Gamma index passing rate with different gamma criteria (global and local 1%/1mm, and local 2%/2mm) with varying low dose thresholds (10%, 40%, and 90%) was also performed using Python PyMedPhys package. No significant dose differences were observed between p-CT def and s-CT. The largest dosimetric differences were observed in air regions, since the algorithm does not detect air cavities inside the body, assigning these voxels a mass density corresponding to soft tissue or fat. Differences were also observed in the external contour and in the boundary of structures due to image registration uncertainties. Figure 1 shows the relative difference of dosimetric parameters for PTV, CTV and OARs. Dosimetric differences for CTV were less than 1%. Higher differences were found in PTV D 98% and OARs. OARs differences were caused by anatomical changes in bladder and rectum as a consequence of time intervals of up to one hour between p-CT and MR scans for some patients. These differences were not completely corrected by the deformable algorithm used. The differences in PTV D 98% were caused by differences in CTV contouring between p-CT and s-CT data sets. Volume differences ranged from 2% to 30%, with larger volumes delineated in p-CT images. Results: