Erschienen in:
16.04.2016 | Vascular-Interventional
Evaluation of a metal artifact reduction algorithm applied to post-interventional flat detector CT in comparison to pre-treatment CT in patients with acute subarachnoid haemorrhage
verfasst von:
Angelika Mennecke, Stanislav Svergun, Bernhard Scholz, Kevin Royalty, Arnd Dörfler, Tobias Struffert
Erschienen in:
European Radiology
|
Ausgabe 1/2017
Einloggen, um Zugang zu erhalten
Abstract
Objectives
Metal artefacts can impair accurate diagnosis of haemorrhage using flat detector CT (FD-CT), especially after aneurysm coiling. Within this work we evaluate a prototype metal artefact reduction algorithm by comparison of the artefact-reduced and the non-artefact-reduced FD-CT images to pre-treatment FD-CT and multi-slice CT images.
Methods
Twenty-five patients with acute aneurysmal subarachnoid haemorrhage (SAH) were selected retrospectively. FD-CT and multi-slice CT before endovascular treatment as well as FD-CT data sets after treatment were available for all patients. The algorithm was applied to post-treatment FD-CT. The effect of the algorithm was evaluated utilizing the pre-post concordance of a modified Fisher score, a subjective image quality assessment, the range of the Hounsfield units within three ROIs, and the pre-post slice-wise Pearson correlation.
Results
The pre-post concordance of the modified Fisher score, the subjective image quality, and the pre-post correlation of the ranges of the Hounsfield units were significantly higher for artefact-reduced than for non-artefact-reduced images. Within the metal-affected slices, the pre-post slice-wise Pearson correlation coefficient was higher for artefact-reduced than for non-artefact-reduced images.
Conclusion
The overall diagnostic quality of the artefact-reduced images was improved and reached the level of the pre-interventional FD-CT images. The metal-unaffected parts of the image were not modified.
Key Points
• After coiling subarachnoid haemorrhage, metal artefacts seriously reduce FD-CT image quality.
• This new metal artefact reduction algorithm is feasible for flat-detector CT.
• After coiling, MAR is necessary for diagnostic quality of affected slices.
• Slice-wise Pearson correlation is introduced to evaluate improvement of MAR in future studies.
• Metal-unaffected parts of image are not modified by this MAR algorithm.