Active Shape Model based Correlative Assessment of Statistical Parameters to Detect Cataract

Abstract

The lens opacity acquired due to clouding of eye lens causing blindness is called a cataract, its detection is challenge. Based on positional occurrence of opacity in a lens it is categorized as nuclear sclerotic, posterior sub-capsular, and cortical cataract. Excessive use of steroids medicines, smoking, ageing, or eye injury lens cause cataract. Gradual and progressive growth of opacity is occurring in lens due to agglomeration of water and protein molecules. The maximum reflected rays form object, that focusing at retina lead to clear understanding of object image. Opacity inside lens causes refraction, reflection and diffraction of reflected light from object. This affects interpretation of image due to poor retinal reception. The patient may be advised treatment or change in lifestyle to prohibit the further growth of cataract or to improve vision. Researches are suggesting methods for detection of cataract from lens images by analysis of color, structural, and optical parameters. The research work presented uses wide slit illumination lens images for analysis to detect cataract. Due to elliptical shape of lens proposed system prefers active shape model with region properties to segment lens structure. Mean pixel value and uniformity are parameters used to distinguish lens without cataract from lens with cataract. The statistical parameters are dependent on probability of pixel values. Hence, using these parameters, system can be made invariant to source illumination. Correlation is performed between parameter vector of input image, and that of known images from database to identify and classify cataract type.