EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A robust index for metal artifact quantification in computed tomography

Jochen Cammin iD

Journal of Applied Clinical Medical Physics · 2024-08 · 25 (8)

DOI: 10.1002/acm2.14453 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Applied Clinical Medical Physics, online erschienen am 25.06.2024. Verfasst von Jochen Cammin. Beteiligte Einrichtung: Department of Radiation Oncology, University of Maryland, Baltimore, Maryland, USA. Hauptthemen laut MEDLINE: Artifacts, Phantoms, Imaging, Metals, Tomography, X-Ray Computed. Die Arbeit wurde bisher 10-mal zitiert (Crossref) und nennt 26 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Objective assessment of metal artifact strength and the effectiveness of metal artifact reduction algorithms in computed tomography requires a quantitative metric. Metrics described in the literature are typically employed to compare the artifact strength in images reconstructed from the same raw data, but their robustness against varying scan conditions and repeated scans over time as it occurs in periodic quality assurance has not been investigated.

Purpose

A new robust metric for quantifying metal artifacts in computed‐tomography images is proposed and compared to other commonly used metrics.

Methods

The proposed artifact metric is based on the location parameter of the Gumbel distribution, described previously in the literature, but normalized to the location parameter in a background region‐of‐interest to obtain a noise‐independent artifact metric. The metric was compared to three other quantitative metal artifact metrics (artifact‐index, contrast‐to‐noise ratio, Gumbel‐evaluation method) by evaluating metals artifacts in phantom scans and in clinical images. Robustness of the artifact metrics was evaluated using repeated scans with varying noise and against small variations in the selected regions‐of‐interest.

Results

The proposed artifact metric was independent of the underlying image noise and could be reproduced more consistently under slight changes of the region‐of‐interest within the artifact than the other investigated methods. The coefficient‐of‐variation was 5.7% on average with varying regions‐of‐interest in phantom scans and 2.5% in patient scans compared to 9.2% in phantoms scans and 9.9% in patient scans for the next‐best performing noise‐independent metric. Setup reproducibility was better than 5% and was comparable to the other metrics. The new metric correlated linearly with the artifact strength. The contrast‐to‐noise ratio, although often used in artifact quantification, was found to be an inadequate metric due to its lack of robustness against minute changes in the position, size, and pixel values of the region‐of‐interest chosen for calculating the metric and because it showed no correlation with the artifact strength.

Conclusions

A new metal artifact metric has been proposed that is robust under changing scan conditions and less sensitive to user‐dependent choices of the region‐of‐interest than other metrics. The new metric is straightforward to calculate and simple to implement in software commonly used for evaluation of medical imaging systems.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

ArtifactsPhantoms, ImagingMetalsTomography, X-Ray ComputedAlgorithmsImage Processing, Computer-AssistedHumansSignal-To-Noise Ratio

Stichwörter der Autor:innen

Gumbel distributionartifact indexcomputed tomographymetal artifactsquality assurance

Substanzen

Metals

Autor:innen und Institutionen

  • Department of Radiation Oncology, University of Maryland, Baltimore, Maryland, USA.

ORCID-Profile: Jochen Cammin

Zitieren

APA

Cammin, J. (2024). A robust index for metal artifact quantification in computed tomography. Journal of Applied Clinical Medical Physics, 25(8). https://doi.org/10.1002/acm2.14453

Vancouver

Cammin J. A robust index for metal artifact quantification in computed tomography. Journal of Applied Clinical Medical Physics. 2024;25(8). doi:10.1002/acm2.14453

BibTeX
@article{Cammin2024,
  author = {Cammin, Jochen},
  title = {A robust index for metal artifact quantification in computed tomography},
  journal = {Journal of Applied Clinical Medical Physics},
  year = {2024},
  volume = {25},
  number = {8},
  doi = {10.1002/acm2.14453}
}
Kostenlos starten →