ESTRO 2023 - Abstract Book

S746

Monday 15 May 2023

ESTRO 2023

Results Accuracy of the MIL model on TR data was 86% and 74% for error type and magnitude classification, respectively; outperforming a CNN trained on TI data, which scored 79% and 62%. Additionally, step 2 of the MIL model allows some insight into which segments/linac angles are important for the model to identify the correct errors. Conclusion MIL is an efficient method for identifying treatment errors in large-scale TR dose verification data. Besides being computationally efficient, it improves error identification accuracy compared to CNNs for TI data, and has the potential to provide insight into model decisions. References 1. Schyns et al. 2016 Phys Med Biol PD-0900 Treating patients with a cardiac implant on a 1.5T MR-linac: a multi-centre experience R. Keesman 1 , E. van der Bijl 1 , L. Kerkmeijer 1 , N. Tyagi 2 , O. Akdag 3 , J. Wolthaus 3 , S. van de Pol 3 , J. Noteboom 3 , M. Intven 3 , M. Fast 3 , A. van Lier 3 1 Radboud University Medical Center, Department of Radiation Oncology, Nijmegen, The Netherlands; 2 Memorial Sloan- Kettering Cancer Center, Department of Medical Physics, New York, USA; 3 University Medical Center Utrecht, Department of Radiotherapy, Utrecht, The Netherlands Purpose or Objective Patients with a cardiac implantable electronic device (CIED) might need MR-guided radiotherapy (MRgRT). Guidelines exist to mitigate CIED-associated risks for MRI examinations and radiotherapy, separately. MRgRT introduces two additional risks, particularly for thoracic cancer patients, namely, distortion of the electrocardiogram (ECG) signal due to MRI and degradation of image fidelity due to presence of metals. We present a clinical workflow that includes methods to assess these risks and share our clinical experience treating 7 CIED patients using MRgRT. Materials and Methods Existing CIED protocols for MRI and conventional radiotherapy were adapted locally to create an MRgRT protocol for CIED patients at three different institutions (Fig. 1). Treatment slots were extended with 15 minutes to facilitate cardiac monitoring tasks. Eligible patients were included in this study. An ECG was acquired with an Invivo Expression MR400 (Philips Healthcare) for a healthy volunteer during baseline and while acquiring motion monitoring MRI for pelvis, pancreas, and lung on a Unity 1.5T MR-linac (Elekta AB). The efficacy of a B0 field-mapping procedure, to assess geometric fidelity of the MRI, was verified for moving objects. To this end, phantom (Quasar MRI4D, ModusQA Inc.) measurements simulating a moving lung target near a static CIED or CIED leads in (cardio)respiratory movement were used to compare field maps acquired in various static positions with the in situ dynamic situation.