Numerical evaluation of high-energy, laser-Compton x-ray sources for contrast enhancement and dose reduction in clinical imaging via gadolinium-based K-edge subtraction

Abstract

Conventional x-ray sources for medical imaging utilize bremsstrahlung radiation. These sources generate large bandwidth (BW) x-ray spectra with large fractions of photons that impart a dose, but do not contribute to image production. X-ray sources based on laser-Compton scattering can have inherently small energy BWs and can be tuned to low dose-imparting energies, allowing them to take advantage of atomic K-edge contrast enhancement. This paper investigates the use of gadolinium-based K-edge subtraction imaging in the context of mammography using a laser-Compton source through simulations quantifying contrast and dose in such imaging systems as a function of laser-Compton source parameters. Our simulations indicate that a K-edge subtraction image generated with a 0.5% BW (FWHM) laser-Compton x-ray source can obtain an equal contrast to a bremsstrahlung image with only 3% of the dose.

Bayesian optimization of laser-Compton x-ray sources for medical imaging applications

“…The electron beam was simulated using General Particle Tracer (GPT) code. This beam was simulated to mimic the expected output of an 11.424 GHz X-band linear accelerator (LINAC) that is currently under construction at Lumitron Technologies, Inc. [10]. This electron beam consists of a micro-bunch charge of 25 pC, mean electron energy between 25 MeV and 100 MeV, electron energy spread of 0.03%, beam emittance of 0.2 mm mrad, electron bunch length of 2 ps and a root-mean-square (RMS) spot size of 5 microns.…”

“…X-ray propagation through a computational phantom was computed using a Beer-Lambert method described previously [10]. This method is fast decreasing computation time and is accurate to established Monte Carlo code within 0.15% accuracy in signal and 1% in noise.…”

“…From here, the image is analyzed by taking the contrast-to-noise ratio (CNR) of the respective inserts. These values are fed to a Bayesian optimization routine that evaluates the following objective function modified from what was previously used for this specific problem [10]. This Bayesian routine then returns an estimate of the electron beam Lorentz factor, 𝛾, and other optimizable parameters to generate a new spectrum for x-ray propagation.…”

“…The computational phantom was a 4.2 cm thick mammography phantom that has been used in a previous gadolinium study [10]. The phantom was designed to accurately simulate K-edge subtraction mammography.…”

A New X-ray Medical-Image-Enhancement Method Based on Multiscale Shannon–Cosine Wavelet

“…The enhancement of medical images is a task of high practical value. In fact, many current medical images, especially X-ray DR images of low-dose projection data, are often blurred in the original image [ 1 , 2 ]. In medicine, these blurred images contain a lot of important details and information that is crucial for medical diagnoses.…”