High-low reflectivity enhancement based retinal vessel projection for SD-OCT images.

Abstract

The retinal vessel visualization from spectral-domain optical coherence tomography (SD-OCT) images is important for ocular disease diagnosis and multimodal retinal image processing. The purpose is to display the retinal vessel in a single projection image from 3D SD-OCT images by using the light absorption and shadow characteristics of the retinal vessel. The authors present a novel retinal vessel projection method for SD-OCT images, which utilizes the light absorption and shadow characteristics of the retinal vessel, called high-low reflectivity enhancement (HLRE) method. The reflectivity of the retinal vessel increases between the internal limiting membrane and inner nuclear layer-outer plexiform layer (INL-OPL) layers because of the light absorption, and the reflectivity below the retinal vessel decreases because of the influence of the retinal vessel shadow. A retinal vessel mask image generated based on the reflectivity characteristics of the retinal vessel is used to enhance the subvolume projection image restricted between the INL-OPL and Bruch's membrane layers. Experimental results with 22 SD-OCT cubes from 12 patients and 10 normal persons demonstrate that the authors' method is more effective in displaying the retinal vessel than the summed-voxel projection and other five region restriction based projection methods. The average of the mean difference between the retinal vessel and background regions based on their HLRE method is 0.1921. The proposed HLRE method was more effective for the visualization of the retinal vessels than the state-of-art methods because it provides higher contrast and distinction.