Publication date: December 2018
Source: Academic Radiology, Volume 25, Issue 12
Author(s): Takashi Ueguchi, Ryota Ogihara, Sachiko Yamada
Rationale and Objectives
To investigate the accuracy of dual-energy virtual monochromatic computed tomography (CT) numbers obtained by two typical hardware and software implementations: the single-source projection-based method and the dual-source image-based method.
Materials and Methods
A phantom with different tissue equivalent inserts was scanned with both single-source and dual-source scanners. A fast kVp-switching feature was used on the single-source scanner, whereas a tin filter was used on the dual-source scanner. Virtual monochromatic CT images of the phantom at energy levels of 60, 100, and 140 keV were obtained by both projection-based (on the single-source scanner) and image-based (on the dual-source scanner) methods. The accuracy of virtual monochromatic CT numbers for all inserts was assessed by comparing measured values to their corresponding true values. Linear regression analysis was performed to evaluate the dependency of measured CT numbers on tissue attenuation, method, and their interaction.
Results
Root mean square values of systematic error over all inserts at 60, 100, and 140 keV were approximately 53, 21, and 29 Hounsfield unit (HU) with the single-source projection-based method, and 46, 7, and 6 HU with the dual-source image-based method, respectively. Linear regression analysis revealed that the interaction between the attenuation and the method had a statistically significant effect on the measured CT numbers at 100 and 140 keV.
Conclusions
There were attenuation-, method-, and energy level-dependent systematic errors in the measured virtual monochromatic CT numbers. CT number reproducibility was comparable between the two scanners, and CT numbers had better accuracy with the dual-source image-based method at 100 and 140 keV.
from Imaging via a.sfakia on Inoreader https://ift.tt/2BtS5VA
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.