Fourier analysis of digital retinal images in estimation of cataract severity

Abstract

To evaluate the efficacy of Fourier analysis of digital retinal images in grading cataract severity. A total of 330 eyes from 165 patients were divided into 287 phakic and 43 pseudophakic eyes. All patients underwent clinical examination, colour fundus photography and photoscreening. Phakic eyes were classified into 232 cataractous and 55 control eyes using the Lens Opacity Classification System (LOCS III). Assessment was repeated at 6 weeks postoperatively for 157 eyes in the cataractous group and in 24 pseudophakic eyes. The predictive value in assessing cataract severity using discrete Fourier transforms in fundus image analysis was compared with LogMAR visual acuity and LOCS III grading. The mean age of the population was 71.7 years +/- 13.4 (range 19-92) there were 70 (42.4%) males and 95 (57.6%) females. The mean LogMAR visual acuity was 0.44 +/- 0.41 (range -0.2 to 1.4). In the cataract group there were 232 eyes (80.8%), the mean age was 75.5 years +/- 9.3 (range 21-92). Fourier analysis of retinal images showed a moderate correlation with LOCS III (R(2) = 0.59) in grading cataract severity and demonstrated a comparable correlation with visual acuity (R(2) = 0.39) as LOCS III (R(2) = 0.44). As with LOCS III score the Fourier value had a poor correlation with visual acuity in clear (R(2) = 0.05) and pseudophakic lenses (R(2) = 0.07). Fourier analysis is a useful automated method in grading of cataract severity for both pure and mixed types of cataract; however, similar to many objective systems it cannot determine the anatomic type of cataract.