Study of attenuation correction in sparse-view tomographic dose reconstruction with a static multi-stack scintillating fiber detector

Abstract

High-resolution two-dimensional dosimetry is important for radiotherapy quality assurance. Scintillating fiber tomodosimetry is attractive because of its near water equivalence and fine spatial sampling, but most implementations rely on detector rotation. Here, we report Monte Carlo simulations of a static multi-stack scintillating fiber detector for sparse-view 2D dose reconstruction and study the attenuation correction required for multi-depth acquisition. TOPAS Monte Carlo simulations were performed for a 6 MV photon beam in water. Rotating acquisitions with 19 and 10 projections were first used to validate sparse-view reconstruction for an I-shaped field. A static geometry with 10 stacked fiber arrays was then evaluated. Direct reconstruction of the raw stacked sinogram produced strong distortion due to depth-dependent attenuation. Three correction methods were compared: percent-depth-dose (PDD) scaling, Beer–Lambert fitting, and robust adaptive normalization. For the rotating configuration, reconstructions with 19 and 10 projections agreed well with the TOPAS reference, with gamma pass rates of 99% and 95%, respectively. For the static configuration, reconstruction without attenuation correction yielded a gamma pass rate of 55.5%. After correction, PDD scaling gave the best agreement with the reference (γ pass rate 93.1%), slightly better than Beer–Lambert scaling. These results support rotation-free sparse-view scintillating fiber dosimetry, provided that depth-dependent attenuation is corrected. Future work will focus on experimental validation and extension of the stacked architecture toward quasi-three-dimensional dosimetry.