Scatter Removal in Photon-Counting Dual-Energy Chest X-Ray Imaging Using a Moving Block Method: A Simulation Phantom Study

TL;DR

This study shows that a moving block method effectively reduces scatter artifacts in photon-counting chest X-rays, improving image quality without increasing radiation dose.

Contribution

The study introduces a dose-neutral scatter correction method for photon-counting dual-energy chest imaging using a moving block approach.

Findings

Scatter correction reduced image artifacts and restored contrast in dual-energy subtraction images.

The moving block method achieved a 0.4% average pixel-wise error in scatter estimation.

Contrast improvements were most significant in larger phantoms and aluminum-enhanced images.

Abstract

This work investigates the impact of scatter correction on photon-counting dual-energy chest radiography using a moving block method, focusing on quantifying improvements with the IEC 62220-2-1 dual-energy metrics. A modified LucAl-based chest phantom with PMMA and aluminum inserts was modeled in three sizes (small, standard, large) to represent different patient sizes. Monte Carlo simulations with MC-GPU and the Photon Counting Toolkit were used to simulate a CdTe photon-counting detector with two energy thresholds at 30 and 70 keV. Scatter was estimated from blocker shadows at 25 positions, interpolated across the field of view, and smoothed with a Gaussian filter (σ=5.0 mm), then subtracted separately from low- and high-energy images. Performance was evaluated using the per-feature dual-energy contrast (DEC) and the kerma-normalized dual-energy subtraction efficiency (DSE) with all acquisitions normalized to an entrance air kerma of 1 mGy to reflect clinical exposure conditions. In simulations, the moving block estimate reproduced the true scatter distribution with an average pixel-wise error of 0.4%. Scatter contamination introduced visible artifacts in the dual-energy subtraction images, particularly in aluminum-enhanced (Al-enhanced) images, and reduced contrast for target materials by up to 25%, as reflected in both DEC and DSE values at a fixed dose. Scatter correction restored image contrast, increased DEC for target materials while keeping non-target DEC low, and reduced edge artifacts across phantom sizes with the largest gains in the large phantom. These results support the moving block method as a dose-neutral strategy to improve dual-energy subtraction performance in photon-counting chest radiography.

Figures