Scholarly record

THE REPRODUCIBILITY OF BLACK-WHITE AREA RATIO CORRESPONDING TO THE ECHOGENICITY GRADE ANALYSIS OF MEDIA TEMPORAL LOBE BRAIN STRUCTURE IN B-IMAGE TO ALZHEIMER DISEASE DIAGNOSIS

Jiří Blahuta

Abstract

The presented paper is focused on digital image processing of medial temporal lobe (MTL) brain structure displayed in B-image. The analysis of the echogenicity grade of the MTL is an important part of Alzheimer?s Disease diagnosis. The image processing is based on the computation of the ratio of black/white pixels in square-shape Region of Interest (ROI) in which the MTL is visible. This ratio could correspond with the MTL echogenicity grade. For this analysis, the core of our developed computer application created for evaluation of substantia nigra echogenicity grade is used. It is a pilot study to evaluate the reproducibility of the ratio between two non-experienced observers with validation by an experienced sonographer. For this study, in total 97 B-images have been analysed. Majority of the images from the set have two different resolutions; original acquired images and a magnified segment with the MTL. There are two key goals of the study ? the reproducibility of the ratio between two observers and to prove that the ratio is not significantly different in the case of original image in comparison with a magnified segment. The set of 97 B-images was analysed four times by two independent observers with achieved correlation coefficient r = 0.69. In addition, maximum, minimum and mean differences were computed for reproducibility assessment. From the obtained results, the reproducibility of the ratio can be considered as a well-reproducible value between independent non-experienced observers and between B-image in normal size and the magnified segment; the difference of the ratio between the same ROI in normal size in the magnified segment does not exceed 5 %. The results have been discussed with an experienced sonographer who verified our results with his professional analysis. In the future, the ratio can be a useful value to determine the MTL echogenicity grade. For this purpose, the percent error between two independent observers have been calculated and also between two measurements from the same observer. The error does not exceed 28 % between 2 non-experienced observers and does not exceed 12 % between the measurement from the same observer.

Publication Impact Profile

• Captures
• Mendeley - Readers: 4

PLUMX see details

DOI resolver: doi.org/10.5593/sgem2019/2.1/s07.073

Publication details

Back to publication list