Radioluminescence 3D dose reconstruction for FLASH and conventional radiotherapy

FLASH radiotherapy requires dosimetric techniques capable of operating at ultra-high dose rates (UHDR), where conventional detectors often suffer from saturation, limited temporal resolution or lack of spatial information. Radioluminescence imaging (RLI) in plastic scintillators represents a promising alternative, enabling real-time and high-resolution visualization of dose deposition. In this work, we present a proof-of-concept methodology for three-dimensional (3D) dose reconstruction from radioluminescence images, exploiting cylindrical beam symmetry and inverse tomographic reconstruction. Three reconstruction algorithms (Fourier-based Abel inversion (FBAI), filtered back-projection (FBP) and maximum-likelihood expectation maximization (MLEM)) were first evaluated using Monte Carlo (MC) simulations to assess reconstruction accuracy against ground-truth dose distributions. Experimental validation was then performed using a plastic scintillator cube irradiated under UHDR conditions with a 9 MeV FLASH electron linac, as well as under conventional dose rate using a 6 MeV electron linac and a 6 MeV photon CyberKnife system. Radioluminescence images were acquired with a laterally positioned CMOS camera and reconstructed dose distributions were compared with Gafchromic (GC) film measurements using gamma index analysis and depth-dose profiles. All reconstruction methods successfully reproduced the global depth-dose behavior. Experimental results demonstrated good agreement with GC film dosimetry, with gamma values predominantly below unity and reconstructed profiles closely following reference measurements across different dose rates and field sizes. These findings demonstrate that radioluminescence-based 3D dose reconstruction is a viable and experimentally simple approach for dosimetry in both conventional and FLASH radiotherapy, offering a valuable complementary tool for beam monitoring and quality assurance in emerging UHDR applications.