Home Publications Research Grants Inventions & Patents Awards Additional Research Activities Faculties & Institutions Research Areas

An automatic algorithm measuring the retinal intercapillary area to assess diabetic retinopathy

Schottenhamml J, Moult EM, Ploner S, Lee B, Lu CD, Husvogt L, Waheed NK, Duker JS, Hornegger J, Fujimoto JG (2016)

---

Publication Language: English

Publication Type: Conference contribution, Abstract of a poster

Publication year: 2016

Book Volume: 57

Pages Range: 5960

Edition: 12

Conference Proceedings Title: Investigative Ophthalmology & Visual Science

Event location: Seattle, WA, USA

URI: https://www5.informatik.uni-erlangen.de/Forschung/Publikationen/2016/Schottenhamml16-AAA.pdf

Abstract

Purpose : Several studies demonstrated the utility of optical coherence tomography angiography (OCTA) in evaluating vascular abnormalities in patients with diabetic retinopathy (DR). Thus far, automatic quantitative analysis has focused on measurements of the foveal avascular zone (FAZ) and capillary density. Analysis of the intercapillary areas (ICA) of the perifoveum using manual segmentation has also been demonstrated. The purpose of this study is to develop a fully automated algorithm for quantifying retinal perfusion impairment using ICA displayed on OCTA images. The longer term objective is to develop sensitive and robust metrics for the diagnosis of DR and to quantify DR progression.

Methods : A 1050 nm, 400 kHz swept-source (SS)-OCT system was used to perform OCTA imaging of 89 eyes from 51 diabetic patients and 63 eyes from 32 normal subjects. Of the 89 eyes from diabetic patients, 51 eyes had no clinically detected retinopathy, 29 eyes had non-proliferative diabetic retinopathy (NPDR), and 9 eyes had proliferative diabetic retinopathy (PDR). OCTA en face images of the retinal vasculature were generated. The vessel structure in each image was highlighted using a vesselness filter; next, a binary image was obtained through a marching algorithm designed to exploit continuity constraints. Finally, the ICA were determined via connected components and subsequently used to generate false color images highlighting areas of non-perfusion.

Results : Representative ICA maps for a normal eye and a NPDR eye are shown in Figure 1. The ICA maps can visualize areas of non-perfusion and enable rapid assessment of the vascular changes that occur in DR. Similar, but less pronounced changes were observed in eyes of diabetics without retinopathy.

Conclusions : A fully automatic algorithm for quantifying ICA in OCTA images was developed. Preliminary results suggest ICA may be a useful metric for assessing diabetes and DR. Further work is needed to develop quantitative measures, validate the algorithm, and evaluate robustness in the presence of OCTA image artifacts.

Authors with CRIS profile

Involved external institutions

How to cite

APA:

Schottenhamml, J., Moult, E.M., Ploner, S., Lee, B., Lu, C.D., Husvogt, L.,... Fujimoto, J.G. (2016, September). An automatic algorithm measuring the retinal intercapillary area to assess diabetic retinopathy. Poster presentation at ARVO Annual Meeting 2016, Seattle, WA, USA, US.

MLA:

Schottenhamml, Julia, et al. "An automatic algorithm measuring the retinal intercapillary area to assess diabetic retinopathy." Presented at ARVO Annual Meeting 2016, Seattle, WA, USA 2016.

BibTeX: Download