Click here to

Session: X-Ray-Based Breast Imaging [Return to Session]

A Workflow to Improve Dual-Energy Spectral Separation in Direct-Indirect Dual-Layer Detector for Contrast-Enhanced Breast X-Ray Imaging

X Duan*, H Huang, W Zhao, SUNY Stony Brook, East Setauket, NY


TU-E-201-7 (Tuesday, 7/12/2022) 11:00 AM - 12:15 PM [Eastern Time (GMT-4)]

Room 201

Purpose: In contrast-enhanced breast imaging, using a dual-layer detector can obtain low-energy (LE) and high-energy (HE) images simultaneously without the deleterious effect of patient motion. We proposed a direct (a-Se) - indirect (CsI) dual-layer detector with k-edge filter for spectral breast imaging. LE images with dual-layer technique are inevitably contaminated with HE photons, which decreases lesion conspicuity in dual-energy (DE) subtracted images. We aim to improve energy spectral separation in dual-layer technique by minimizing the effect of HE photon contamination in the LE image.

Methods: We conducted Monte Carlo simulations on a digital phantom with adipose and glandular tissue, solid masses, cysts, and iodinated masses. Ag k-edge filter was used to form spectrum with two energy peaks below and above the k-edge of iodine. The x-rays incident on the back layer (BL) has similar beam quality as the HE portion of x-rays interacted in the front layer (FL). A virtual front HE component was estimated from the BL image and denoised with a deep neural-network. The denoised virtual front HE image was subtracted from the FL image to generate the virtual LE image.

Results: The contrast between adipose tissue and other tissue types in the LE image with the dual-layer technique was ~60% of that in regular mammography. Virtual LE images improved contrast to ~87% of that in regular mammography, by minimizing the effect of HE photon contamination. Iodinated mass SDNR squared per dose increased after using the virtual LE image instead of FL image for DE subtraction (140%, 126%, 108%, 92% and 76% for iodinated mass size of 1, 2, 3, 4 and 5 cm, with 2 mg/ml iodine concentration).

Conclusion: We developed a workflow to minimize the contamination of HE photon in FL LE images, which improved the image contrast and increased iodinated masses detectability.


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email