Background: Manual segmentation of the Foveal Avascular Zone (FAZ) has a high level of variability. Research into retinas needs coherent segmentation sets with low variability.
Methods: Retinal optical coherence tomography angiography (OCTA) images from type-1 diabetes mellitus (DM1), type-2 diabetes mellitus (DM2) and healthy patients were included. Superficial (SCP) and deep (DCP) capillary plexus FAZs were manually segmented by different observers. After comparing the results, a new criterion was established to reduce variability in the segmentations. The FAZ area and acircularity were also studied.
Results: The new segmentation criterion produces smaller areas (closer to the real FAZ) with lower variability than the different criteria of the explorers in both plexuses for the three groups. This was particularly noticeable for the DM2 group with damaged retinas. The acircularity values were also slightly reduced with the final criterion in all groups. The FAZ areas with lower values showed slightly higher acircularity values. We also have a consistent and coherent set of segmentations with which to continue our research.
Conclusions: Manual segmentations of FAZ are generally carried out with little attention to the consistency of the measurements. A novel criterion for segmenting the FAZ allows segmentations made by different observers to be more similar.
Keywords: deep capillary plexus; diabetes mellitus; diabetic retinopathy; foveal avascular zone; optical coherence tomography angiography; superficial capillary plexus.