Ain Shams Engineering Journal (Dec 2024)
Advanced shape detection in Optical Coherence Tomography (OCT) imaging
Abstract
Automated detection of morphological changes in retinal layers using Optical Coherence Tomography (OCT) imaging assists optometrists in making accurate decisions regarding eye diseases. While OCT devices can clearly show these changes, they often lack precise descriptions supported by calculations and numerical comparisons to the normal state of the retina. Therefore, this work proposes and applies a novel technique which applied on 423 OCT images, enabling a thorough examination of these morphological variations. The technique begins with image restoration using a Wiener filter, followed by segmentation using Modified Fuzzy C-means technique to obtain a clear binary image. Shape descriptor parameters are then extracted from the binary OCT image. Detection of morphological alterations is assessed by identifying changes in these shape parameters. The results reveal significant changes in retinal layer shapes across various diseases. For example, in diabetes, there is a 19% increase in minor length of retinal structure, while the major length changes by 21%. Additionally, solidity and elongation show a 21% shift. Notably, diabetic disease exhibits the highest number of holes, with a 70% occurrence compared to other retinal diseases. In contrast, Central Serous Chorioretinopathy (CSC) shows a 7% decrease in major length, while Macular Hole (MH) disease records a notable 17% increase in eccentricity. As a future direction, using the extracted descriptive information to train an artificial neural network could automate diagnosis procedures, enhancing the efficiency and accuracy of disease detection and management.
