Suppression of the Eyelash Artifact in Ultra-Widefield Retinal Images

• Online Access: http://hdl.handle.net/2047/d10008883

Retinal imaging is used by optometrists and ophthalmologists to screen for and diagnose eye and non-eye diseases. It is believed that indicators of systemic diseases and precursors to different eye conditions often exhibit first on the periphery of the retina. Current retinal examination methods provide a narrow field of view (about 5%) and can therefore miss eye and non-eye disorders that may be present and can be detected on the periphery of the retina. However as the field of view becomes greater, certain artifacts can be captured in the image making its analysis more challenging. An example of this is the presence of eyelashes. In this work, an automated image-pair registration method known as the Generalized Dual-Bootstrap Iterative Closest Point (GDB-ICP) algorithm [2]-[4], was used to suppress the eyelash artifact, of ultra-widefield retinal images. The percent of suppression evidenced in the mosaic created by the algorithm was quantified. For the pair of images used in this work the percent of suppresion obtained, was 6.13% in regards to the overall image.