Multivendor Spectral-Domain Optical Coherence Tomography Dataset, Observer Annotation Performance Evaluation, and Standardized Evaluation Framework for Intraretinal Cystoid Fluid Segmentation.

Jing Wu,Ana-Maria Philip,Georg Langs,Ursula M Schmidt-Erfurth,Sebastian M Waldstein,Christian Simader,Dominika Podkowinski,Bianca S Gerendas

doi:10.1155/2016/3898750

Abstract

Development of image analysis and machine learning methods for segmentation of clinically significant pathology in retinal spectral-domain optical coherence tomography (SD-OCT), used in disease detection and prediction, is limited due to the availability of expertly annotated reference data. Retinal segmentation methods use datasets that either are not publicly available, come from only one device, or use different evaluation methodologies making them difficult to compare. Thus we present and evaluate a multiple expert annotated reference dataset for the problem of intraretinal cystoid fluid (IRF) segmentation, a key indicator in exudative macular disease. In addition, a standardized framework for segmentation accuracy evaluation, applicable to other pathological structures, is presented. Integral to this work is the dataset used which must be fit for purpose for IRF segmentation algorithm training and testing. We describe here a multivendor dataset comprised of 30 scans. Each OCT scan for system training has been annotated by multiple graders using a proprietary system. Evaluation of the intergrader annotations shows a good correlation, thus making the reproducibly annotated scans suitable for the training and validation of image processing and machine learning based segmentation methods. The dataset will be made publicly available in the form of a segmentation Grand Challenge.

Highlights

Spectral-domain optical coherence tomography (SD-OCT) is the most important ancillary test for the diagnosis of sight degrading diseases such as retinal vein occlusion (RVO), agerelated macular degeneration (AMD), and glaucoma [1]
The detection of intraretinal cystoid fluid (IRF) is a important indicator of disease severity and change in exudative macular disease as increased retinal thickness has shown to correlate with poor visual acuity [6]; automated detection and segmentation methods are required to employ “big data” in visual acuity and treatment progression prediction
IRFs have been chosen as the basis for this multivendor reference dataset and grader performance assessment [7, 8]

Summary

Introduction

Spectral-domain optical coherence tomography (SD-OCT) is the most important ancillary test for the diagnosis of sight degrading diseases such as retinal vein occlusion (RVO), agerelated macular degeneration (AMD), and glaucoma [1]. SDOCT is a noninvasive modality for acquiring high resolution, 3D cross-sectional volumetric images of the retina and the subretinal layers, in addition to retinal pathology such as intraretinal fluid, subretinal fluid, and pigment epithelial detachment [2, 3]. Detection and segmentation of such pathologies are an important step in the diagnosis of disease severity and treatment success, as well as an early stage towards the accurate prediction of both [4, 5]. The detection of intraretinal cystoid fluid (IRF) is a important indicator of disease severity and change in exudative macular disease as increased retinal thickness has shown to correlate with poor visual acuity [6]; automated detection and segmentation methods are required to employ “big data” in visual acuity and treatment progression prediction. IRFs have been chosen as the basis for this multivendor reference dataset and grader performance assessment [7, 8].

Objectives

Methods

Results

Conclusion