Room: Exhibit Hall | Forum 9
Purpose: To develop and validate a generalized frequency-dependent weighting scheme forcombining the sub-images of Multi-Layer Imager (MLI) devices for beam's-eye view imaging inradiation therapy.
Methods: A homogeneous MLI consisting of four layers of Gd2O2S:Tb (GOS) scintillators and aheterogenous MLI featuring a combination of a GOS and a novel LKH5 scintillator were simulatedusing the GATE Monte Carlo software. The detector response was characterized in terms of theModulation Transfer Function (MTF), Normalized Noise-Power Spectrum (NNPS) and DetectiveQuantum Efficiency (DQE). A frequency-dependent weighting factor for each individual layer wasderived such that the total DQE would be maximized across all spatial modes. The final optimizedimage is obtained as a weighted linear combination of the sub-images from the individual layers.For validating the proposed optimization scheme, resulting changes in the MTF, NNPS and DQEwere evaluated for the simulated MLI detector configurations.
Results: The optimal weighting factors that maximize the DQE of an MLI configuration were foundto be proportional to the ratio between the MTF and NNPS of each individual layer. For thehomogenous MLI configuration, the weighting approach yielded a marginally lower NNPScompared to the unweighted reference metric. Since the MTF remains unchanged, a DQE(0)improvement of 0.4% for the optimal combination was obtained. In contrast, for the heterogenousMLI configuration, the method translated into an observable improvement in the MTF and the NNPSacross all different frequencies resulting in a 1.0% enhanced DQE(0).
Conclusion: An analytical approach was used to develop a frequency-dependent weightingscheme for the optimal combination of the images using MLIs for applications in radiotherapy.The proposed method can enhance the quality of the final image across spatial modes, especiallywhen layers of different properties are combined, potentially enabling a new class ofheterogeneous MLIs.
Not Applicable / None Entered.