Digital imaging and microtexture analysis of the nerve fiber layer

Purpose: To describe the design of a digital retinal nerve fiber layer (RNF L) imaging techniques and present a new approach to measure the differences in RNFL patterns. Methods: A digital camera body is connected to a wide-angle camera to obtai n images of the RNFL, which are displayed in workstations throughout the cl inic. In the on-line archive, images in Joint Photographies Experts Group ( JPEG) format (100 KB per frame) are used. The hypothesis that changes in RN FL structure can be seen as changes in the microtexture of digital images w as tested using an information theoretical approach (Kullback Information D istance, KID). A large KID value indicates a large difference, and a small KID value indicates a small difference in microtexture between the two regi ons. The material of this pilot study consists of 9 patients with glaucoma, 8 patients with ocular hypertension, and 7 normal subjects. Results: The median KID value in the glaucoma group was 3.5, compared with the median KID values of 0.6 in the control groups. Although a trend could be seen in the measured values, because of a small sample size, the differe nces were not statistically significant. Five of 24 (21%) KID values overla pped between the glaucomatous group and the other two groups. Conclusion: Although digital imaging produces good quality RNFL images, fur ther research is needed to establish minimum accepted specifications for di gital imaging. In this pilot study, only the microtexture of the RNFL was m easured in digital images. In the future, the approach can be expanded to i nclude also properties of macrotexture and full color palette.