Abstract
The Foveal Avascular Zone (FAZ) is a capillary-free area that is placed inside the macula and its morphology and size represent important biomarkers to detect different ocular pathologies such as diabetic retinopathy, impaired vision or retinal vein occlusion. Therefore, an adequate and precise segmentation of the FAZ presents a high clinical interest. About to this, Angiography by Optical Coherence Tomography (OCT-A) is a non-invasive imaging technique that allows the expert to visualize the vascular and avascular foveal zone. In this work, we present a robust methodology composed of three stages to model, localize, and segment the FAZ in OCT-A images. The first stage is addressed to generate two FAZ normality models: superficial and deep plexus. The second one uses the FAZ model as a template to localize the FAZ center. Finally, in the third stage, an adaptive binarization is proposed to segment the entire FAZ region. A method based on this methodology was implemented and validated in two OCT-A image subsets, presenting the second subset more challenging pathological conditions than the first. We obtained localization success rates of 100% and 96% in the first and second subsets, respectively, considering a success if the obtained FAZ center is inside the FAZ area segmented by an expert clinician. Complementary, the Dice score and other indexes (Jaccard index and Hausdorff distance) are used to measure the segmentation quality, obtaining competitive average values in the first subset: 0.84± 0.01 (expert 1) and 0.85± 0.01 (expert 2). The average Dice score obtained in the second subset was also acceptable (0.70± 0.17), even though the segmentation process is more complex in this case.
Highlights
Optical Coherence Tomography Angiography (OCT-A) [1] represents a new image modality that is commonly used in the ophthalmic field and allows the expert to visualize non-invasively the vascularization at different levels of the retinal layers
We compare our results to the manual labeling of each expert clinician and to the results obtained in [20], considering the different levels of complexity associated with the six possible configuration criteria
We are more interested in the behavior of the localization and segmentation method in each category of difficulty, comparing the results obtained with those labeled by the expert
Summary
Optical Coherence Tomography Angiography (OCT-A) [1] represents a new image modality that is commonly used in the ophthalmic field and allows the expert to visualize non-invasively the vascularization at different levels of the retinal layers. In step 1, it is necessary to extract several OCT images in the region of interest by forming a cube of OCT images. This process is repeated several times, being obtained different OCT cubes at different. Once the differences are calculated, we obtain a new OCT cube where the blood vessels are highlighted in each section (step 2). By selecting the desired layer (superficial or deep), which is specified by the configuration of the OCT-A capture device, the cross-section of the cube is made in the region of interest, being the final OCT-A image obtained (step 3). OCT-A images are inspected by the expert clinicians in two different depth configurations: superficial and deep plexus; both of them representing the more relevant regions
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
