FAZSeg

FAZSeg segments and quantifies the Foveal Avascular Zone (FAZ) in retinal OCT angiography images to measure superficial and deep FAZ dimensions for ophthalmological research.

Key Features:

• Algorithm: Implements the New Automated Software Method (NAM) for automated FAZ segmentation and quantification.
• Image input specifications: Processes 6mm × 6mm retinal images acquired with the Angioplex protocol on a Cirrus 5000 device (Carl Zeiss Meditec Inc.) centered on the fovea at 420×420 pixels.
• Layer-specific segmentation: Produces separate segmentations and measurements for superficial and deep FAZ layers.
• Validation: Segmentation outputs are validated against manual segmentations with reported performance metrics.
• Performance metrics: NAM achieves 91.40% overall accuracy; superficial layer DSC 0.94 and SSIM 0.97; deep layer DSC 0.96 and SSIM 0.98.
• Validation cohort: Performance reported on 93 healthy subjects (30 emmetropic, 63 myopic) with age ranges and mean ages provided in the study.

Scientific Applications:

• Ophthalmological research: Quantifies FAZ dimensions to study the impact of ocular diseases on retinal microvasculature.
• Myopia studies: Measures FAZ alterations associated with high myopia.
• Anatomical and clinical assessment: Enables layer-specific FAZ analysis for anatomical studies and clinical monitoring of retinal health.

Methodology:

FAZSeg employs the New Automated Software Method (NAM) to analyze FAZ dimensions and processes 6mm × 6mm Angioplex OCTA images from a Cirrus 5000 (Carl Zeiss Meditec Inc.) centered on the fovea at 420×420 pixels, with validation against manual segmentations.