CNR enhancement in breast implant phantoms via spectral x-ray imaging

The limited visibility of breast lesions underlying silicone implants hampers the early detection of breast cancer-like features, highlighting the need for imaging methods that improve contrast behind implants. The present study investigates, both computationally and experimentally, the feasibility of combining spectrum optimization and material decomposition to enhance the contrast-to-noise ratio (CNR) of micro-calcifications in mammographic phantoms with breast implants. The goal is to reconstruct an image with improved CNR by using recovered energy-independent maps of two reference materials and applying the Beer-Lambert equation to the energy bin with the highest identified CNR. The computational study evaluates the approach using aluminum oxide and hydroxyapatite micro-calcifications embedded in standard-density breast tissue and partially obscured by silicone breast implants of four different thicknesses. Energy-bin images are generated from a simulated monochromatic x-ray source and weighted using four Tungsten polychromatic spectra with an additional filter: aluminum (1 mm) at three voltages and rhodium (0.05 mm) at 28 kVp. The experimental study independently assesses the same approach using aluminum oxide micro-calcifications within the Mammo-156 phantom, partially covered by a 250 cc compressed silicone implant. Polymethyl methacrylate slabs are added to simulate a 5 cm-thick breast. The experimental setup consists of a Tungsten micro-focus x-ray source and a hybrid photon counting detector Timepix3, which enables the acquisition of energy bin images and the estimation of entrance radiation dose. Exposure time is adjusted to achieve an entrance radiation dose of approximately 4 mGy for each spectrum, which is a typical value used in mammography. Simulation results demonstrate an enhancement in CNR in the reconstructed images compared with both the input energy-bin images with the highest CNR and the energy-weighted images, for both types of micro-calcifications and in both regions-inside and outside the implant-across all implant thicknesses. Experimental results also show the enhancement in CNR in the reconstructed images under realistic imaging conditions, for micro-calcifications in both inside and outside the implant region.