Researchers at Duke University have recently established a foundation for the use of virtual imaging trials in effective assessment and optimization of CT and radiography acquisitions and analysis tools to help manage the coronavirus disease (COVID-19) pandemic. The research was published in an open-access article in the Journal of Roentgenology.
Virtual imaging trials have two main components—representative models of targeted subjects and realistic models of imaging scanners—and the authors of this AJR article developed the first computational models of patients with COVID-19, while showing, as proof of principle, how they can be combined with imaging simulators for COVID-19 imaging studies.
“For the body habitus of the models,” lead author Ehsan Abadi explained, “we used the 4D extended cardiac-torso (XCAT) model that was developed at Duke University.”
Abadi and his Duke colleagues then segmented the morphologic features of COVID-19 abnormalities from 20 CT images of patients with multidiagnostic confirmation of SARS-CoV-2 infection and incorporated them into XCAT models.
“Within a given disease area, the texture and material of the lung parenchyma in the XCAT were modified to match the properties observed in the clinical images,” Abadi et al. continued.
Example of 4D XCAT Phantom Developed at Duke. Top: Representative computational model shows lung stroma intraorgan structure of XCAT phantom that was developed using anatomically informed mathematic model. Inset shows enlarged view for better visibility of details and small structures. Bottom: Voxelized rendition (ground truth) of XCAT phantom highlights detailed model of lung parenchyma. Inset shows enlarged view for better visibility of details and small structures.
Using a specific CT scanner (Definition Flash, Siemens Healthineers) and validated radiography simulator (DukeSim) to help illustrate utility, the team virtually imaged three developed COVID-19 computational phantoms.
Simulated CT Images of COVID-19 on 4D XCAT Phantom. Images of same phantom show simulated CT at 50 (A), 25 (B), and 5 (C) mAs, as well as simulated chest radiograph (D).
“Subjectively, the simulated abnormalities were realistic in terms of shape and texture,” said the authors, adding their preliminary results showed that the contrast-to-noise ratios in the abnormal regions were 1.6, 3.0, and 3.6 for 5-, 25-, and 50-mAs images, respectively.Back To Top
Virtual Imaging Trials Optimize CT, Radiography for COVID-19. Appl Radiol.