ESTRO 2021 Abstract Book

S408

ESTRO 2021

Conclusion Our study is the first attempt to generate virtual biopsy from raw MR without the need of paired data. A new gold standard for heterogeneity of tumor microenvironment thus becomes available to the radiotherapist towards better delineation, characterization of hypoxic or lymphocyte-empty areas and ultimately locally- adaptative dose optimization. Additional features are in development to produce significant clinical value: improvement of resolution until histology granularity, generalization to other locations with challenges on new tissues, and addition of an automatic tumor segmentation pipeline to the virtual biopsy. Integration intro treatment planning in terms of delineation for head and neck cases is ongoing. OC-0523 Cardiac-Synchronized T1 mapping on the Unity MR-linac to guide STereotactic Arrhythmia Radioablation S. Mandija 1 , O. Akdag 1 , A. van Lier 1 , P. Borman 1 , T. Schakel 1 , E. Alberts 2 , O. van der Heide 1 , R. Hassink 3 , M. Alings 3 , J. Verhoeff 1 , F. Mohamed Hoesein 4 , M. Fast 1 1 UMC Utrecht, Radiotherapy, Utrecht, The Netherlands; 2 Philips Healthcare, Philips Healthcare, Best, The Netherlands; 3 UMC Utrecht, Cardiology, Utrecht, The Netherlands; 4 UMC Utrecht, Radiology, Utrecht, The Netherlands Purpose or Objective STereotactic Arrhythmia Radioablation (STAR), based on a single stereotactic radiotherapy fraction (25Gy to VT substrate) is a promising treatment for ventricular tachycardia (VT). STAR is noninvasive and allows access to locations that are unreachable for conventional catheter ablation. To enable STAR, accurate visualization of the VT substrate (scar) is essential. Late-Gadolinium-Enhanced (LGE) MRI is used to image scars in diagnostic MRI. However, while gadolinium can be administered during MRI simulation (MR-sim), it cannot currently be used on the MR-linac because of its potential interaction with radiation. A promising alternative for scar visualization is cardiac T1 mapping, which can differentiate healthy/scar tissue without contrast agent administration (1,2). Here, we investigate the feasibility of cardiac-synchronized MRI on the 1.5T Unity MR-linac (Elekta AB, SE), and quantify the reproducibility and robustness of cardiac T1 mapping compared to a 1.5T Ingenia MR-sim T1 maps were acquired twice on MR-sim and MR-linac on a phantom (Diagnostic Sonar, SC) and 5 healthy volunteers to investigate robustness and reproducibility within/across scanners using a MOdified Look-Locker Inversion recovery (MOLLI 5(3)3) technique with inversion prepulse, cardiac triggering and breath-hold (FOV=30x30x4.8cm 3 , resolution=2x2x8mm 3 , slices=3; gap=1.2cm; TR=2.5ms; TE=1.1ms; SENSE=2, flip- angle=35 o ). The obtained T1 maps were compared with gold-standard inversion recovery single spin echo measurements. Cardiac synchronization hardware (PPU: MRI compatible peripheral pulse-oximeter unit), standardly used on MR-sim, was enabled on the MR-linac for the first time (research mode). Segmentation of regions of interest and statistics were derived using in-house software. Results Cardiac synchronization hardware was successfully installed on the MR-linac to enable cardiac-triggered T1 mapping. T1 measurements (mean±SD) on the phantom are reproducible within/across MR-sim and MR-linac ( Figure1 ). The well-known, small underestimation of T1 values obtained with MOLLI with respect to gold-standard measured T1 values is in line with literature (3). In-vivo T1 values are reproducible across intra/inter-scanner measurements ( Figure2 , septum: blue ROI; blood: red ROI), and agree with literature (1,2). (Philips Healthcare, NL). Materials and Methods