Room: Track 2
Purpose: Image quality in portal imaging can be enhanced by heterogeneous multi-layer imagers (MLI). This work presents the validation of a frequency-based method for combining the sub-images from the different layers.
Methods: A MLI configuration was modeled within the GATE Monte Carlo package by stacking two mega-voltage imager models: a standard Gd2O2S:Tb (GOS) phosphor layer and a novel continuous layer of LKH-5 glass scintillator material. Using two clinical photon beams of 2.5 MV and 6 MV, the modulation transfer function (MTF), noise power spectra (NNPS) and the detective quantum efficiency (DQE) were computed. A frequency-dependent weighting factor was then analytically derived for each layer that maximizes the total DQE of the combined image of all layers. The performance of the proposed weighting scheme was evaluated by quantifying the spatial resolution, contrast-to-noise ratio (CNR) and signal-to-noise ratio (SNR) of simulated planar images of a line-pair phantom and Las Vegas MV imaging phantom.
Results: The proposed method enhances the DQE(0) for the combined detector by 2-3% compared to a straight summation of the images, validating the benefit of the weighting scheme. While the simulated line-pair phantom exhibits slightly lower MTF values for spatial frequencies greater than 0.2 lp/mm compared to the unweighted image, CNR and SNR are improved by around 20%. For the Las Vegas phantom, the weighting improved the CNR by around 30% depending on the evaluated contrast region and the SNR by a factor of 2.5.
Conclusion: An analytic approach was used to develop a frequency-dependent weighting scheme for the optimal combination of the sub-images from novel heterogeneous MLI devices for applications in radiotherapy. The method was validated using Monte Carlo simulations and enhances the quality of the final image in terms of the CNR and SNR across the entire frequency domain independently of the energy of the photon beam.
Funding Support, Disclosures, and Conflict of Interest: This project was partially supported by award No. R01CA188446 from the National Cancer Institute.