Characterization and Monte Carlo Modelling of ElectronFlash LINAC at Institute Curie

The electron LINAC ElectronFlash installed at the Institut Curie (France) is dedicated to the investigation of the FLASH effect on pre-clinical trials and radiobiology studies [1]. The accurate beam characterization is essential for a proper dose calibration. Although a few dosimeters operate under FLASH conditions, e.g., diamond [2], ultra-thin ionization chambers (UTIC) [3], and SiC-based detectors [4], Monte Carlo (MC) simulations are a complementary tool for estimating doses in scenarios where those dosimeters can have a limited use. This study aims to present the experimental and simulated characterization of the ElectronFlash Linac for both FLASH and conventional modalities using a set of dosimeters and MC simulations. Irradiations were performed at 7 MeV, with 0.5 – 5 μs pulse widths, and up to 250 Hz with different poly methyl methacrylate (PMMA) applicators. The experimental setup consisted in a PTW Water Phantom and a PEEK plastic collimator designed for irradiating mouse lungs (Fig. 1, left). The absolute dosimetry was made with Radiochromic EBT3 films, a PTW FlashDiamond, and UTIC to characterize the percentage depth dose (PDD) in that water tank (Fig. 2, left). The scatter factors were evaluated with new SiC detectors designed and fabricated at IMB-CNM. The beam FWHMs were also quantified with the radiochromic films. We used the open-source GATE MC (vs 10.0.2) to model the beamline starting from the experimental data gathered using the dosimeters mentioned above. Simulations were run in the computational cluster TIRANT v4. Considering the experimental work, the PDDs obtained showed loss of electron equilibrium at the beginning of the curve and non-negligible differences between FLASH and conventional modalities (Fig 1, right). Additionally, there was a significant difference (43%) in the FWHMs between these two modalities, which suggests notable differences in the beam configuration for each one. The scatter factors evaluated with SiC diodes were in good agreement with those from flashDiamond. Interestingly, these PDDs yielded an increase of electron energy compared to the previous PDD reported [3], which reinforce the need to model the LINAC. Regarding the MC study, the simulated PDD showed an excellent agreement with the experimental measured (Fig 2, right) validating the accuracy of the MC modelling. This enables calculate the dose distributions in cases where the dosimeters cannot easily be used, e.g., for small field sizes or large dose gradients. This work shows the first comparison of experimental dose distributions with those obtained with accuracy MC model of the electronFLASH LINAC at Inst. Curie.