ESTRO 2024 - Abstract Book

S5949

RTT - Treatment planning, OAR and target definitions

ESTRO 2024

729

Digital Poster

Development of an automatic deformable registration workflow for multi-modal prostate imaging

Andrea Lastrucci 1 , Eva Serventi 2 , Giulio Francolini 3 , Yannick Wandael 1 , Gabriele Simontacchi 3 , Renzo Ricci 1

1 Azienda Ospedaliera Universitaria Careggi, Department of Allied Health Professions, Florence, Italy. 2 Azienda USL Toscana Centro, Department of Allied Health Professions, Prato, Italy. 3 Azienda Ospedaliera Universitaria Careggi, Radiation Oncology Unit, Florence, Italy

Purpose/Objective:

Radiotherapy is one of the curative therapeutic options for the treatment of prostate cancer. Images obtained through planning Computed Tomography (CT) are used to contour the target volume and organs at risk. At the level of the prostate, there are several anatomical structures characterised by small differences in density, often difficult to define on CT. Magnetic resonance images (MRI) are considered the gold standard to differentiate the interface between soft tissues and correctly determine the volume of the prostate. There are numerous studies in the literature confirming the usefulness of CT-MRI multimodal fusion for determining prostate volume. At Prato Radiation Therapy, a customised automated workflow on the MIM Maestro software was developed by the RTTs in collaboration with the multidisciplinary team to perform elastic multimodal registration using the multimodality algorithm, aiming to avoid any intra- and inter-observer variability. To evaluate the correctness of the processing of the obtained image data, a comparison was made between the defined prostate volume in the images from CT, in the MRI obtained with a T2-weighted turbo spin-echo sequence, and finally in MRI (MRIDEF) deformed by the workflow on a recruited population.

Material/Methods:

This retrospective observational study was conducted at a single centre and included 25 patients undergoing radical radiotherapy for prostate cancer. For each patient, prostate volume was delineated on MRI, MRIDEF and CT images, respectively. The aim of the study was to compare the difference in millilitres and percentage volume of the delineated prostate contour in CT and MRIDEF with a gold standard defined by MRI-based delineation.

Results:

The data on the changes in millilitres and the percentage of contour volumes in the three image datasets can be found in Table 1. The comparison of MRI and CT volumes showed that the CT defined prostate volume is overestimated by about 30% on average. Furthermore, a statistically significant difference was observed between the volumes contoured in CT and MRI (p-value < 0.05). On comparing the prostatic volumes outlined in MRI and MRIDEF, the data overlap, and there is no statistically significant difference between the two distributions (p-value > 0.05).