Title:Dual-energy Computed Tomography Imaging from Contrast-enhanced Single-energy Computed Tomography

Abstract:In a standard computed tomography (CT) image, pixels having the same Hounsfield Units (HU) can correspond to different materials and it is therefore challenging to differentiate and quantify materials. Dual-energy CT (DECT) is desirable to differentiate multiple materials, but DECT scanners are not widely available as single-energy CT (SECT) scanners. Here we purpose a deep learning approach to perform DECT imaging by using standard SECT data. We designed a predenoising and difference learning mechanism to generate DECT images from SECT data. The performance of the deep learning-based DECT approach was studied using images from patients who received contrast-enhanced abdomen DECT scan with a popular DE application: virtual non-contrast (VNC) imaging and contrast quantification. Clinically relevant metrics were used for quantitative assessment. The absolute HU difference between the predicted and original high-energy CT images are 1.3 HU, 1.6 HU, 1.8 HU and 1.3 HU for the ROIs on aorta, liver, spine and stomach, respectively. The aorta iodine quantification difference between iodine maps obtained from the original and deep learning DECT images is smaller than 1.0\%, and the noise levels in the material images have been reduced by more than 7-folds for the latter. This study demonstrates that highly accurate DECT imaging with single low-energy data is achievable by using a deep learning approach. The proposed method allows us to obtain high-quality DECT images without paying the overhead of conventional hardware-based DECT solutions and thus leads to a new paradigm of spectral CT imaging.