Abstract
Purpose: :
To demonstrate the new technique for the registration of high–density OCT images to fundus landmarks in fast spectral–domain OCT. To demonstrate the application of fast spectral–domain OCT in studying the behaviors of the parapapillary retinal blood vessel parameters in normal and glaucoma eyes.
Methods: :
High density OCT images can significantly improve the image quality and enhance the visualization of especially the structure of small retinal pathologies. Another advantage of spectral–domain OCT is the generation of en face OCT fundus images, which provide perfect registration for the sectional OCT images. Unfortunately, it is impractical to generate a useful fundus image from the high density OCT scans. However, spatial registration for a high–density OCT image can be achieved by using the OCT fundus image generated from the normal–density scans that cover the area of interest. Three steps were taken: (1) feature detection; (2) feature matching; (3) transformation and spatial location determination. All three steps were performed automatically with computer. High–density circular scans around the optic disc were taken for the study of the parapapillary blood vessel parameters. By using the same technique with generating OCT fundus images, the distribution of retinal reflectivity on the scanned circle was calculated. The axes of the blood vessels were first detected through a series of signal processing. The raw profile of each blood vessel was then fitted to a model of Gaussian distribution centered at the axis of the blood vessel. Geometric parameters of each blood vessel like the diameter can then be extracted from the fitted distribution.
Results: :
Both normal and diseased human eyes were studied with our new techniques. Precise spatial registration was achieved for high quality high–density OCT images (for example 2048 A–scans) on the fundus images generated from normal OCT scans (for example 128x512 A–scans). Preliminary studies on the parapapillary retinal blood vessel diameters for normal and glaucoma eyes showed a decrease of the total blood vessel diameters for glaucoma patients.
Conclusions: :
Spatial registration is very important for quantitative ophthalmic imaging and information extraction. We have developed a technique for precise spatial registration of high–density OCT sectional images on OCT fundus image. This technique solved the registration problem of high–density OCT images. The new technique for blood vessel studies opened a new application area for spectral–domain OCT and has potential application in glaucoma study and diagnosis.
Keywords: imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • image processing • optic disc