Bone-enhanced high contrast X-ray images derived from attenuation estimation related to ultra-low energy X-rays – An application of an energy-resolving photon-counting detector (ERPCD)

Purpose: X-ray diagnosis in medicine is often used for bone diagnosis based on qualitative observation analysis. However, there are often cases where the contrast of bones is reduced because of the existence of soft-tissues, making it difficult to accurately diagnose the bone conditions. Although the algorithm for bone extraction images was proposed using an energy-resolving photon-counting detector (ERPCD), this algorithm can depict “one” bone material (such as hydroxyapatite under the assumption), and it is difficult to adequately depict other components. The purpose of this study is to develop an algorithm for bone-enhanced high-contrast images that can be virtually represented by the attenuation of extremely low-energy X-rays without making any special assumptions. Methods: High-contrast images were virtually generated based on the attenuation rate of ultra-low energy X-rays. It was determined by fitting the mass attenuation coefficient (μ/ρ) curve to the X-ray attenuation values (μt values) measured at middle (30–40 keV) and high (40–60 keV) energy windows, and extrapolating the μt values to those for the low energy region (E = 5–20 keV). When performing the extrapolation, the effective atomic number (Z_eff) of the object was taken into consideration. The methodology was validated by simulating X-ray projections using a digital human body phantom. The frequency of correspondence between the pixel values in the high-contrast image and the Z_eff image was analyzed for each pixel. Results: We succeeded in creating virtual high-contrast X-ray images that reflect the image contrast of monochromatic X-rays of 5–20 keV. It was confirmed that the pixel values in the high-contrast image corresponding to an Z_eff = 7.5 (soft-tissue) were completely separated from those corresponding to an Z_eff = 9 (bone). The optimization of the energy related to the high contrast images was performed based on the contrast-to-noise ratio (CNR) analysis. The high contrast image with 10 keV showed a good CNR value. Conclusions: Based on the analysis of the attenuation information of middle and high-energy X-rays measured by ERPCDs, we succeeded in creating a novel algorithm that can generate a virtual monochromatic image with high contrast.