Session Title: Free Paper Session 25: Imaging IV
Session Date/Time: Sunday 10/09/2017 | 10:00-11:30
Paper Time: 10:24
Venue: Room 113
First Author: : C.Enders GERMANY
Co Author(s): : J. Werner M. Loidl G. Lang G. Lang
To study quantity and quality of image artefacts in optical coherence tomography angiography (OCT-A) in subjects with and without macular disease.
Prospective monocentre study at the outpatient department of an university eye hospital.
OCT-A examination was performed with the spectral domain OCT Cirrus 5000 equipped with the AngioPlex™ module (Carl Zeiss Meditec, Inc., Dublin, USA). Two experienced ophthalmologists who were masked to the results of each other independently analysed the OCT-A-Scans of 50 consecutive patients for segmentation errors, projection and other artefacts (displacement, banding, blink, masking, unmasking, out-of-window, vessel doubling, stretching, and crisscross artefacts) in different segmentations (superficial capillary plexus, deep capillary plexus, avascular layer, choriocapillary layer and choroidal layer). Additionally the general assessability of the OCT-A-images was graded. All OCT-A-images were assessed by two graders. We enrolled healthy controls (6) and patients with either diabetic retinopathy (11), retinal artery occlusion (10), retinal vein occlusion (10), and age-related macular degeneration (13).
Segmentation errors were found in 11 eyes in the superficial capillary plexus, in 20 eyes in the deep capillary plexus, in 21 eyes in the avascular slab, in 11 eyes in the choriocapillary slab and in 8 eyes in the choroidal slab. Positive projection was found in all eyes, mainly in the deep capillary plexus. Negative projection was found in 49 eyes, mainly in the choriocapillary and choroidal slabs. Displacement artefacts were found in 23 eyes. Banding was present in 23 eyes. Blink artefacts were found in 18 eyes, masking in 17 eyes, and unmasking in 1 eye. Out-of-window artefacts were present in 4 eyes. Vessel doubling, stretching and crisscross artefacts were not found. 37 eyes showed good, 12 eyes impaired image assessability, 1 eye was not sufficiently assessable.
All imaging techniques exhibit different kinds of artefacts. Failure to recognize image artefacts may lead to image misinterpretation. As shown in this study, OCT-A image artefacts are common but nevertheless images enabled good to sufficient data interpretation of vascular flow in 49/50 eyes. Segmentation errors of different source occur in a relatively large number of subjects without decreasing general assessability and they can be manually corrected. One of the most important are projection artefacts, that occured in all subjects in any structure below the vessels. Some artefacts occur in a small numbers of patients, some artefacts described in the literature were not identified in our study. Artefacts of OCT-A images are caused either by the method, data processing or cooperation of the patient during image capture. Recognition of artefacts is important for correct image interpretation and decision making.