|
KMID : 1147120130190010031
|
|
Journal of the Korean Society of Imaging Informatics in Medicine
2013 Volume.19 No. 1 p.31 ~ p.37
|
|
|
An Automated Method for Measurement of Noise Level in CT Images
|
|
Chun Min-Soo
Kim Jong-Hyo
|
|
|
Abstract
|
|
|
|
Measurement of the noise level in CT images is an important task in assessing the image quality for protocol optimization and quality control in CT imaging. Despite the importance of appropriate noise evaluation, relatively little studied is the subject on automated measurement of noise level in CT images. The noise measurement is usually carried out by manually drawing a region of interest, and the standard deviation within the region is being used to represent the noise level. While this manual measurement apparently appears to be straightforward and reliable, its reliability in terms of intra- and inter- rater variability has not been fully assessed and thus potentially suffers rater dependent variability. In this study, we have developed a technique for automatically localizing the homogeneous regions by applying the histogram analysis on the sum of gradient within the region of interest, and thereby quantifying the noise level in CT image. We applied the developed technique to 50 CT images of 10 patients (each individual for 5 images) and evaluated its performance in comparison with raters’manual drawing method. Our developed technique has successfully performed the operation in all the test images, and has shown an agreement of 0.87 with the reference measurement which was obtained by pooling of all raters’measurements. Our developed technique has a potential to be used for the assessment of image quality of CT images in various studies evaluating imaging performance of evolving CT techniques.
|
|
|
KEYWORD
|
|
|
Computed Tomography, Noise level quantification
|
|
|
FullTexts / Linksout information
|
|
|
|
|
|
Listed journal information
|
|
|
|