This approach stands out through its robust handling of peripapillary atrophy and vascular shadows using inverse-intensity weighting, while integrating gradient orientation and intensity terms to refine contours. By combining enface and 3D layer information, it achieves accurate, smooth boundaries for both disc and cup. The energy-minimizing active contour and PDM-based classification ensure automated, reproducible segmentation and diagnosis, offering clinicians precise, quantitative metrics for early glaucoma detection and progression monitoring.
Description
The system begins by converting a 3D spectral-domain OCT volume into a two-dimensional enface map through A-scan intensity averaging and normalization. It then locates the optic disc margin using a modified active contour model: low-intensity regions are detected post–median filtering, a weighted Canny edge map is generated with inverse intensity weights, and an initial circle is fitted via Hough transform. The contour iteratively deforms under combined internal (smoothness) and external (gradient magnitude, orientation, intensity) forces, followed by clustering and smoothing of radial profile points. For cup estimation, the internal limiting membrane and retinal pigment epithelium layers are segmented from A-scan peaks, yielding a 3D ONH surface. Cup edges arise from intersections with an RPE-based reference line or fixed-distance plane, and final margins are refined via median filtering and cubic interpolation. From these contours, the system computes disc, rim, and cup areas and volumes, global and sectoral cup-to-disc and rim-to-disc ratios, plus vertical/horizontal C/D ratios, and classifies the ONH by comparing energy fits in point distribution models of normal and abnormal shapes.
Applications
Automated glaucoma screening device
SD-OCT diagnostic imaging software
Ophthalmic clinic decision support
Teleophthalmology remote ONH analysis
Clinical trial imaging endpoint
Advantages
Objective, quantitative measurement of optic nerve head parameters (disc/cup areas, volumes, ratios)
Early and accurate detection of glaucoma through 3D SD-OCT–based analysis
Automated, consistent disc and cup margin segmentation despite peripapillary atrophy or vessel occlusion
Robust smoothing and interpolation ensure high repeatability and reduced operator variability
Comprehensive ONH modeling enables sectoral analysis for localized pathology assessment
Automated normal vs. abnormal classification using point distribution models
Time-saving workflow by eliminating manual delineation and subjective grading
Facilitates monitoring of disease progression with standardized longitudinal measurements
IP Status
Copyright
