Image-Based Dual Energy CT using Optimized Precorrection Functions: A Practical New Approach to Material Decomposition in the Image Domain

摘要

Dual energy CT (DECT) measures the object of interest using two different x-ray spectra in order to provide energy-selective CT images or in order to get the material decomposition of the object. Today, two decomposition techniques are known. Image—based DECT treats the rawdata sets as being independent until they are reconstructed. Prior to reconstruction the rawdata undergoes the typical water precorrection. The reconstructed images are then linearly combined to obtain material-selective DECT images. The second decomposition technique is raw data-based. Rawdata-based DECT is passing the rawdata through a decomposition function D(q_1,p_2) followed by image reconstruction. Rawdata-based decomposition is exact and the final image will not show beam hardening artifacts. However, ra-wadata-based decomposition requires consistent rawdata sets. Our new imaged-hased method uses generalized precorrection functions p_1(q_1) that do not aim at delivering cupping artifact-free single energy images but that aim at delivering an improved DECT image by using the linear combination p_1(q_1)+p_2(q_2), which can also be carried out in image domain. This work compares the ability of the three methods to perform material decomposition and to provide energy-selective monochromatic CT images. Due to its increased degrees of freedom the generalized precorrection functions significantly outperform the image quality achievable with conventional image-based DECT decomposition. Nevertheless, the artifact content is still higher than when using rawdata-based decomposition techniques. Since our method can be realized as a polynomial function or as a look-up table, it can easily be used to substitute the water precorrection functions built into today's scanners. Thereby the approach is a practicable way to improve image-based DECT without changing the scanner software or hardware.