Evaluation of Segmentation of the Superficial and Deep Vascular Layers of the Retina by Optical Coherence Tomography Angiography Instruments in Normal Eyes
News in ophthalmology : Evaluation of Segmentation of the Superficial and Deep Vascular Layers of the Retina by Optical Coherence Tomography Angiography Instruments in Normal Eyes
JAMA Ophthalmol. 2017 Mar 1;135(3):259-262. doi: 10.1001/jamaophthalmol.2016.5327.
Spaide RF1, Curcio CA2. Author information
Abstract
Importance: Correct attribution of vascular features in optical coherence tomography (OCT) angiography depends on accurate segmentation of retinal layers.
Objective: To evaluate the segmentation of retinal layers among 3 OCT angiography instruments in the central macula, an area where the superficial and deep vascular plexuses terminate.
Design, Setting, and Participants: A retrospective review of a representative OCT angiogram from 1 patient and an evaluation of the vascular pattern in an autopsied eye were conducted at a community retina practice at a university laboratory. A set of 3?×?3-mm scans centered on the fovea using the Cirrus 5000, RTVue XR Avanti, and Triton DRI OCT platforms with default layer segmentations were used to evaluate segmentation accuracy of a normal macula of a white man in his 60s as an emblematic example. A representative histologic section from the central macula of a normal eye was used as an exemplar.
Main Outcomes and Measures: Retinal layer segmentation and resultant vascular image compared with vessels as seen in histologic section.
Results: The segmentation slab designed to isolate the superficial vascular plexus included the deep vascular plexus in the central macula for all 3 instruments. None of the instruments produced segmented regions that followed the relevant anatomic layers correctly.
Conclusions and Relevance: Because of inherent errors in segmentation, studies of the superficial and deep vascular plexuses using manufacturer-recommended default settings are likely to be biased. A proposal for an improved segmentation strategy is presented