Purchase this article with an account.
H. Ishikawa, R. Hariprasad, G. Wollstein, L.A. Paunescu, L.L. Price, P.C. Stark, J.G. Fujimoto, J.S. Schuman; New Quality Assessment Parameters for OCT 3 . Invest. Ophthalmol. Vis. Sci. 2003;44(13):3358.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose: To develop a method to evaluate the quality of optical coherence tomography (OCT 3, Carl Zeiss Meditec Inc., Dublin, CA) images in a quantitative and objective fashion. Methods: Raw data files of OCT 3 images (linear macular scan, peripapillary circular scan, and optic nerve head scan) were exported to an IBM compatible PC. A software program of our own design analyzed the files and created histograms for each image. We defined 3 intensity levels on each histogram: Low Reflectivity was defined as the first percentile, Noise was defined as the 75th percentile, and Saturation was defined as the 99th percentile for pixel intensity. An Intensity ratio (IR) was defined as a ratio of Saturation and Low Reflectivity multiplied by 100, and a tissue signal ratio (TSR) was defined as a ratio of pixel population of the higher and the lower half of the range between Saturation and Noise. Finally, a quality index (QI) was defined as a product of IR and TSR. To evaluate the performance of these parameters, the outcome results were compared with subjective 3-level-grading (excellent, acceptable, and poor) that 3 OCT experts categorized. Since the OCT 3 system does not provide signal to noise ratio except for the peripapillary scan, we did not compare it with our parameters. Results: 63 images of 21 subjects (7 each for normal, early, and advanced glaucoma) were enrolled in this study. Subjects were selected in a consecutive and retrospective fashion from our OCT imaging database. All quality parameters showed significant differences among grading categories (all p<0.0001, ANOVA); the QI showed the least overlap between grades (IR>TSR>QI). Discriminating power (both excellent and acceptable grades were treated as acceptable) was highest with QI (area under ROC (receiver operator characteristic) curve 0.91 (IR), 0.96 (TSR), and 0.99 (QI)). Conclusions: A QI may permit automated objective, quantitative assessment of OCT image quality similar to that of an expert human observer.
This PDF is available to Subscribers Only