A novel direct-indirect dual-layer flat-panel detector for contrast-enhanced breast imaging: Monte Carlo simulation

Abstract

Background

Contrast-enhanced digital mammography (CEDM) and contrast-enhanced digital breast tomosynthesis (CEDBT) utilize weighted subtraction of low-energy (LE) and high-energy (HE) images to highlight breast lesions with iodine uptake. Typically, LE and HE images are acquired with two separate exposures, with intervals up to 10 s for CEDM and longer for CEDBT. Patient motion during these intervals can lead to incomplete subtraction of normal breast structures in recombined dual-energy (DE) images. The residual tissue structure masks lesions, distorts lesion margins, and reduces contrast enhancement– thus degrading diagnostic accuracy.

Purpose

This study introduces a novel direct-indirect dual-layer flat-panel detector (DI-DLFPD) to eliminate the effect of patient motion by acquiring LE and HE images simultaneously. The proposed DI-DLFPD system consists of a silver (Ag) x-ray filter at the tube port, a 200 µm direct conversion amorphous selenium (a-Se) front layer (FL) detector, and a 400 µm indirect conversion cesium iodide (CsI) back layer (BL) detector. We validated the DI-DLFPD design using Monte Carlo simulations, assessing iodine objects detectability in CEDM images across different system designs.

Methods

Simulations were generated with a digital phantom mimicking a 4-cm thick breast with 25 % glandularity. Cylindrical iodinated objects with 5-mm size in diameter and height were embedded with 1, 2, 3, and 5 mg/ml iodine concentration. A 5-mm cubic was designed as glandular tissue inside the phantom to evaluate the background cancellation in DE subtracted images. Four imaging scenarios (A–D) were Monte Carlo simulated with different configurations of the x-ray filter and FL and BL detectors. Post-imaging processing, including image registration and modulation transfer function matching, was conducted. Besides, an analytical method for generating virtual FL-LE images was introduced to further improve the iodine detectability in the DE subtracted image. The signal difference to noise ratio (${\mathrm{SDNR}}_{\mathrm{mean}}$) of iodinated objects in DE images served as the figure of merit for quantitative comparison.

Results

The proposed DI-DLFPD configuration outperformed all other scenarios in DE image quality, effectively removing cubic normal glandular tissue and revealing iodinated objects at all concentrations. ${\mathrm{SDNR}}_{\mathrm{mean}}$ is higher in the proposed scenario A than in B, C, and D by 74 %, 19 %, and 88 %, respectively. Incorporation of virtual FL-LE images further improved ${\mathrm{SDNR}}_{\mathrm{mean}}$ by 10 %.

Conclusions

In this study, the great potential of the proposed DI-DLFPD for CE breast imaging was demonstrated with the high iodine objects detectability in Monte Carlo simulated CEDM images. The advantage of generating virtual FL LE images for better energy separation was evidenced by the high iodine objects detectability in DE images. This study not only provides vital guidance for DI-DLFPD system design but also lays the foundation of subsequent physical experiments with a DI-DLFPD prototype.