EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC 4.0

Establishing quality control action limits for CT number accuracy in spectral images using an American College of Radiology phantom

Xinhui Duan, Yue Zhang

Medical Physics · 2023

DOI: 10.1002/mp.16626 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Medical Physics, 2023. Verfasst von Xinhui Duan und 1 weiteren Autor:innen. Beteiligte Einrichtung: Department of Radiology, UT Southwestern Medical Center, Dallas, Texas, USA. Hauptthemen laut MEDLINE: Phantoms, Imaging, Tomography, X-Ray Computed, Societies, Medical. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 28 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0). Gefördert von: Cancer Prevention and Research Institute of Texas.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Computed tomography (CT) number accuracy is important for quality assurance in CT imaging. However, in dual‐energy CT imaging, there are no widely used action limits for CT number accuracy in spectral images, information that is urgently needed.

Purpose

To establish action limits for spectral CT images using longitudinal spectral data and an American College of Radiology (ACR) phantom.

Methods

An ACR accreditation phantom was scanned routinely as part of a quality control program in our institution. We selected and analyzed 57 continuous weekly scans. The CT numbers or the density values of conventional and spectral images, including virtual monoenergetic images (40, 50, 70, 120, and 200 keV), iodine maps, calcium suppressed, and virtual non‐contrast images, were measured in the four inserts (solid water, bone, polyethylene, and acrylic) of the phantom. Longitudinal data were analyzed for correlation using Pearson's correlation coefficient ( r ) and standard deviation (SD). The SD ratios between spectral images and conventional images were calculated and the action limits for spectral images were established based on the action limits from the ACR.

Results

Strong to very strong correlations ( r > 0.70 or r < −0.70) were found among most spectral image types except the 200 keV images using solid water, polyethylene, and acrylic inserts ( r = [−0.45, 0.64]). The SD ratio was highest for the 40 keV images, ranging from 2.8 to 6.5. The action limits of the bone insert were baseline ± 5.3 mg/mL for the iodine map and ranged from baseline ± 23.0 HU to baseline ± 391.9 HU for the other image types. The action limits for solid water ranged from baseline ± 4.1 HU to baseline ± 25.3 HU. The results for the polyethylene and the acrylic insert were close to those for solid water. Baselines can be established using the average of the initial 5∼10 measurements.

Conclusions

Using longitudinal data, we estimated the action limits for CT number accuracy in the spectral images. This paves the way for establishing a comprehensive quality control program for spectral CT imaging.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Phantoms, ImagingTomography, X-Ray ComputedSocieties, MedicalQuality ControlUnited StatesHumans

Stichwörter der Autor:innen

ACR phantomdual-energy CTquality control

Autor:innen und Institutionen

  • Department of Radiology, UT Southwestern Medical Center, Dallas, Texas, USA.

Förderung

  • Cancer Prevention and Research Institute of Texas

Zitieren

APA

Duan, X., & Zhang, Y. (2023). Establishing quality control action limits for CT number accuracy in spectral images using an American College of Radiology phantom. Medical Physics. https://doi.org/10.1002/mp.16626

Vancouver

Duan X, Zhang Y. Establishing quality control action limits for CT number accuracy in spectral images using an American College of Radiology phantom. Medical Physics. 2023. doi:10.1002/mp.16626

BibTeX
@article{Duan2023,
  author = {Duan, Xinhui and Zhang, Yue},
  title = {Establishing quality control action limits for CT number accuracy in spectral images using an American College of Radiology phantom},
  journal = {Medical Physics},
  year = {2023},
  doi = {10.1002/mp.16626}
}
Kostenlos starten →