ESTRO 2024 - Abstract Book

S4910

Physics - Quality assurance and auditing

ESTRO 2024

(Ingenia MR-RT, Philips) on a flat couch top with standard immobilization systems. Initially, 40 patients also received a backup sCT for commissioning purpose and testing. The sCT images were generated using the MRCAT Pelvis algorithm from Philips. Treatment plans were optimized and calculated in Eclipse using the AcurosXB algorithm. Daily kV cone beam CT scans were acquired with TrueBeam (Varian). In addition, a new DL algorithm (a pix2pix GAN, developed and trained in-house) was applied and evaluated to generate sCT from daily CBCT, with the scope to reduce artifacts and improve image quality dose calculation purposes (CBCTsCT). Specific Hounsfield Unit (HU) to mass density curves were used for both CTS and CBCTsCT.

Results:

For the initial cohort of 40 patients, dose comparisons between sCT-CT and sCT-CBCTsCT demonstrated minimal differences in DVH parameters. The mean dose discrepancy to the target was below 0.8% for all cases, and below 0.5% for bladder and rectum. Gamma passing rates (1%-1mm) exceeded 97%, 96.1%, and 94.8% for thresholds of 20%, 50%, and 80%, respectively. Among the remaining patients without backup CT, the mean sCT-CBCTsCT dose differences to the target remained below 0.9% for all patients. Gamma passing rates (1%-1mm) exceeded 96.5%, 95.6%, and 94.5% for thresholds of 20%, 50%, and 80%, respectively.

Conclusion:

This study strongly supports the safe integration and clinical progression of an MRI-only workflow for prostate cancer patients. The incorporation of first-fraction CBCT, complemented by DL correction methods, stands out as a robust QA tool, guaranteeing precise dose delivery within the MRI-Only Workflow. Future efforts are focusing on automating QA procedures using machine learning methods and establishing robust pass/fail criteria.

Keywords: MR-only planning, synthetic CT, CBCT

2584

Digital Poster

A novel method for retrospective evaluation of treatment planning parameters for QA in radiotherapy

Robert Schindhelm, Paul Käthner, Sonja Wegener, Bülent Polat, Gary Razinskas

University Hospital Würzburg, Department of Radiation Oncology, Würzburg, Germany

Purpose/Objective:

Within the treatment planning process, the evaluation of dosimetric values in the current plan is essential to ensure high-quality standards. A standardized and automated evaluation of such parameters significantly streamlines the daily clinical workflow. The software solution ClearCheck by Radformation offers automated verification of dose constraints and plan parameters, among other checks. In general, these analyses are applicable to various entities as well as fractionation schemes for each patient and each plan. The plan-specific evaluations conducted for each patient