Posters

Exploration of automated image analysis of enhanced-depth-imaging optical coherence tomography in patients with central serous chorioretinopathy

Poster Details

First Author: H.Zaki UK

Co Author(s):    R. Chhabra   S. Mahmood   T. Aslam   Z. Ali           

Abstract Details

Purpose:

To develop algorithms for automated analysis of Enhanced Depth Imaging Optical Coherence Tomography scans (EDI-OCT) and explore differences in choroidal thickness between healthy eyes compared to eyes with Central Serous Chorioretinopathy (CSC).

Setting:

Manchester Royal Eye Hospital / University of Manchester

Methods:

A retrospective observational feasibility study conducted from April to September 2017. We developed algorithms using the MatLab platform to segment choroid and assess thickness in a fully automated manner from Heidelberg EDI-OCT images. We studied outcomes from the algorithm in patients with CSC compared to normal patients. In the first part of the study, 5 EDI-OCT sections were included for each subject and average results were used for the measurement of depth. In the second part, only the section with highest level of activity (most hyper-fluorescent by ICGA) was used for analysis and compared to corresponding section in the controls.

Results:

The study included 32 subjects, 16 Acute CSC and 16 controls. Automated algorithms were able to accurately segment out choroidal structures as assessed clinically by retinal experts . Using all 5 sections, the choroidal thickness was not significantly higher in CSC than in controls (P>0.05) . However, when ‘active’ locations were studied, results demonstrated statistically significant greater thickness in patients with CSC (P<0.05 ).

Conclusions:

We produced automated systems that resulted in clinically satisfactory segmentation of choroid from EDI OCT images. For this study, we did not find a noticeable difference in the choroidal thickness when comparing the entire macular regions. Differences were significant if we compared choroid in the actively leaking sections. Published reports also show some inconsistency of choroidal thickness, perhaps due to varying means of measurement including subjective measures and variations in types of imaging and specific methods of image capture. Future studies will investigate choroidal characteristics in more detail using standardised and optimized capture of EDI scans and enhanced automated analyses.

Financial Disclosure:

None