Evaluation of metal artefact reduction using dual-step adaptive thresholding technique in computed tomography
Loading...
Date
2018
Authors
Hassan, Mohd Norsyafi
Journal Title
Journal ISSN
Volume Title
Publisher
Pusat Pengajian Sains Perubatan, Universiti Sains Malaysia
Abstract
Background: Metal objects present in CT images may give rise to streak artefact. In the
presence of severe artefacts, image quality may be extensively degraded and important
clinical findings and pathology in the vicinity of the metal objects may be obscured. The
purpose of this study is to evaluate the effectiveness of the dual-step adaptive thresholding
technique as a method of metal artefact reduction in CT studies.
Methodology: A total of 14 CT studies which contained metal-induced artefacts resulted
from various surgical implants were retrieved from the Picture Archive Communication
System (PACS). The CT images were corrected using the DSAT algorithm in MATLAB
workspace to generate the artefact-corrected images with acceptable quality. Both groups
of original images and artefact-corrected images were evaluated quantitatively using
noise and SNR and qualitatively using visual evaluation by 2 evaluators. Level of
significance was determined (p < 0.05).
Results: A significant reduction of the noise were noticed in the corrected CT images
following DSAT technique for metal artefact correction with the mean noise of 14.576 ±
11.7 as compared to the original images with mean of 40.177 ± 23.785 (p < 0.0005). A
significant improvement of SNR was also demonstrated following DSAT correction with
the mean SNR of 3.877 ± 3.931 for the corrected images in comparison to 3.614 ± 2.839
for the original images (p = 0.017). Visual evaluation has demonstrated reduced
appearance of metal artefacts with increased conspicuity of adjacent structures (p < 0.05).
Conclusion: Metal artefact correction using dual-step adaptive thresholding technique
has the ability to suppress metal-induced artefacts with significant improvement of image
quality.
Description
Keywords
Tomography , X-Ray computed