Semi-Automated Labelling of Cystoid Macular Edema in OCT Scans

Greenwood, Luke (2019) Semi-Automated Labelling of Cystoid Macular Edema in OCT Scans. Masters thesis, University of Lincoln.

Documents
Semi-Automated Labelling of Cystoid Macular Edema in OCT Scans
Thesis
[img]
[Download]
[img]
Preview
PDF
Greenwood, Luke – Computer Science – December 2019.pdf - Whole Document

14MB
Item Type:Thesis (Masters)
Item Status:Live Archive

Abstract

The analysis of retinal Spectral Domain Optical Coherence Tomography (SDOCT) images by trained medical professionals can be used to provide useful insights into various diseases. It is the most popular method of retinal imaging due to its non invasive nature and the useful information it provides for making an accurate diagnosis, however there is a clear lack of publicly available data available to researchers in the domain. In this report, a deep learning approach for automating the segmentation of Cystoid Macular Edema (fluid) in retinal OCT B-Scan images is presented that is consequently used for volumetric analysis of OCT scans. This solution is a fast and accurate semantic segmentation network which makes use of a shortened encoderdecoder UNet-like architecture with an integrated DenseASPP module and Attention Gate for producing an accurate and refined retinal fluid segmentation map. The network is evaluated against both publicly and privately available datasets; on the former the network achieved a Dice coefficient of 0.804, thus making it the current best performing approach on this dataset and on the very small and challenging private dataset, it achieved a score of 0.691. Due to the issue presented by a lack of publicly available data in this domain, a Graphical User Interface that aims to semi-automate the labelling process of OCT images was also created, thus greatly simplifying the process of dataset creation and potentially leading to an increase in labelled data production.

Divisions:College of Science > School of Computer Science
ID Code:47494
Deposited On:07 Dec 2021 11:56

Repository Staff Only: item control page