Abstract:Purpose:To investigate whether deconvolution methods can improve the scatter estimation under different blurring and noise conditions for blocker‐based scatter correction methods for cone‐beam X‐ray computed tomography (CBCT).Methods:An “ideal” projection image with scatter was first simulated for blocker‐based CBCT data acquisition by assuming no blurring effect and no noise. The ideal image was then convolved with long‐tail point spread functions (PSF) with different widths to mimic the blurring effect from the finite focal spot and detector response. Different levels of noise were also added. Three deconvolutionMethods:1) inverse filtering; 2) Wiener; and 3) Richardson‐Lucy, were used to recover the scatter signal in the blocked region. The root mean square error (RMSE) of estimated scatter serves as a quantitative measure for the performance of different methods under different blurring and noise conditions.Results:Due to the blurring effect, the scatter signal in the blocked region is contaminated by the primary signal in the unblocked region. The direct use of the signal in the blocked region to estimate scatter (“direct method”) leads to large RMSE values, which increase with the increased width of PSF and increased noise. The inverse filtering is very sensitive to noise and practically useless. The Wiener and Richardson‐Lucy deconvolution methods significantly improve scatter estimation compared to the direct method. For a typical medium PSF and medium noise condition, both methods (∼20 RMSE) can achieve 4‐fold improvement over the direct method (∼80 RMSE). The Wiener method deals better with large noise and Richardson‐Lucy works better on wide PSF.Conclusion:We investigated several deconvolution methods to recover the scatter signal in the blocked region for blocker‐based scatter correction for CBCT. Our simulation results demonstrate that Wiener and Richardson‐Lucy deconvolution can significantly improve the scatter estimation compared to the direct method.

The Effects of Error Magnitude and Bandwidth Selection for Deconvolution with Unknown Error Distribution.

Deconvolution with Unknown Error Distribution Interpreted as Blind Isotonic Regression

Deconvolution Estimation in Measurement Error Models: the R Package Decon

Identifiability of Multichannel Blind Deconvolution and Nonconvex Regularization Algorithm

Conditional Density Estimation in Measurement Error Problems

Deconvolution Density Estimation with Heteroscedastic Errors Using SIMEX

Uniform confidence bands in deconvolution with unknown error distribution

Multiplicative deconvolution under unknown error distribution

Blind deconvolution algorithm for spectrogram super-resolution restoration

Rapid, Robust, and Reliable Blind Deconvolution Via Nonconvex Optimization

Geophysical Model Enhancement Technique Based on Blind Deconvolution

Deconvolution of repeated measurements corrupted by unknown noise

Maximum Likelihood Deconvolution of Beamforming Images with Signal-Dependent Speckle Fluctuations

Semiparametric Model And Superefficiency In Blind Deconvolution

SU‐E‐I‐08: Investigation of Deconvolution Methods for Blocker‐Based CBCT Scatter Estimation

High-Resolution Power Spectral Estimation Method Using Deconvolution

Bayesian bandwidth estimation for a nonparametric functional regression model with mixed types of regressors and unknown error density

Partial Deconvolution with Inaccurate Blur Kernel

Improving Time Estimation by Blind Deconvolution: with Applications to TOFD and Backscatter Sizing

Iterative Deconvolution for Calibrating Quantum Control Pulses.

Image Reconstruction in Light-Sheet Microscopy: Spatially Varying Deconvolution and Mixed Noise