|
Dual
Energy Decomposition of Alexandra Lauric Basak Ulker Karbeyaz David Rozas Anton Deykoon Ram Naidu Zhengrong Ying Carl Crawford October 15, 2007 This work is motivated by the resurgence of interest in dual-energy applications in medical CT generated by the introduction of the Siemens Definition Dual Source CT Scanner in 2005. The use of dual energy CT for clinical applications has the potential of advancing the use of CT as a combined morphological and functional imaging tool and of redefining the role of CT for soft tissue imaging by providing tools to better differentiate soft tissue detail. Here we investigate how dual-energy may be used to extract additional information from medical CT data. In traditional single-energy CT systems, data is acquired at one X-ray energy and the resulting CT numbers approximate the mass density of the scanned materials. With dual-energy scanning, data is acquired at both low and high X-ray energies (e.g., at 80kVp and 140kVp) and additional material information becomes available. A process known as decomposition is applied to the low and high energy CT data to extract material properties, such us density and atomic number. The decomposition process can be applied to either projection CT data (i.e., scanner data before image reconstruction) or to CT images. We refer to these methods as pre-decomposition and post-decomposition, respectively. We use the two decomposition methods to determine the mass densities and the effective atomic numbers of the materials being scanned. We evaluate the two methods and show preliminary results using simulated data and CT scanner data. The pre-decomposition of the high and low energy X-ray projections is based on the work described by Ying et al. [1]. The post-decomposition of the CT images corresponding to high and low energy scans is based on the work described by Klueg [2]. Both methods were initially developed for use in EDS (explosive detection systems) and we apply these two methods to medical CT data. Our experimental results show that pre-decomposition method is more accurate than post-decomposition in estimating densities and effective atomic numbers. The effective atomic number is defined as the atomic number of a hypothetical single element that would give the same X-ray attenuation as the substance being evaluated assuming the same density. When having access to both projection data and image data, we recommend that pre-decomposition should be preferred to post-decomposition. However, the ability to perform post-decomposition in image space can be an important advantage in situations where one does not have access to projection data and control over offline reconstruction algorithms. We recommend to further investigate the two methods and determine how they compare in terms of additional performance measures such as noise, resolution and artifacts. Our goal in using dual-energy decomposition on medical CT data is to determine if the additional information extracted with dual-energy capability can improve the performance of routine medical image processing tasks, such as segmentation, registration, image analysis and visualization. We consider the work presented here as the first step towards this goal. Further investigation is needed to determine if the combination of mass density and effective atomic number can uniquely identify scanned materials and prove valuable for the image processing tasks listed above. [1] Z. Ying, R. Naidu and C. Crawford, “Dual Energy
Computed Tomography for Explosive Detection,” Journal of X-ray Science and
Technology 14 (2006), pp 235-256. [2] R. E. Klueg, “Quantitative Dual-Energy Computed
Tomography of Small Diameter Specimens,” The 4th International Aviation
Security Technology Symposium, 2006. 1 This
document is part of CIS Technical Report, 07-12, Analogic
Proprietary, October 15, 2007 |