Fourier analysis of circumpapillary retinal nerve fiber layer thickness in optical coherence tomography for differentiating myopia and glaucoma

Abstract

Differentiating glaucoma from myopic eye is a challenge to ophthalmologists. We try to develop a new discrete Fourier transform (DFT) model for analyzing optical coherence tomography data for the circumpapillary retinal nerve fiber layer (cpRNFL), and investigate DFT as a new diagnostic tool for glaucomatous myopic eyes. The thicknesses of 12 equidistant cpRNFL points were transformed into 6 signals in the frequency domain, ranging from 1 to 6 Hz. In all 232 eyes, generalized linear model showed that 1 Hz, 2 Hz, and 4 Hz were associated with glaucoma, high myopia, and the interaction between glaucoma and high myopia. The 3 Hz signal was associated with glaucoma and high myopia exclusively. A receiver operating characteristic curve analysis of the 3 Hz signals showed areas under the curves of 0.93 (95% CI 0.90-0.96) and 0.93 (95% CI 0.88-0.98), for diagnosing glaucoma in all subjects and in the highly myopic group, respectively. The DFT model is useful to differentiate glaucoma from non-glaucomatous change and showed potential as a diagnostic tool for glaucomatous myopic eyes. Myopia is the most common ocular problem in the world. The prevalence of myopia in Asia is higher than that in Africa, Europe, or America. 1-4 Globally, it is estimated that, between the present and the year 2050, the prevalence of myopia and high myopia will continue to grow. 5 Glaucoma is one of the leading causes of blindness. It is characterized as progressive degeneration of retinal ganglion cells and thinning of the circumpapillary retinal nerve fiber layer (cpRNFL), which is the superficial layer of retina in the human eye. In recent years, optical coherence tomography (OCT) has become an important tool for detecting early anatomical changes in the cpRNFL. OCT is currently used for disease screening and progression assessments. However, in high myopic eyes, analyses of OCT images of the cpRNFL have resulted in high false-positive error rates, which makes the diagnosis of glaucoma complicated. 6,7 Clinically, visual field examination, which reveals the patient's optic nerve function, is the gold standard test to diagnose and evaluate the progression of the glaucoma. However, it takes about half an hour to perform the test in the both eyes of a patient with a whole-process accompanied technician. It was a huge burden for patients and medical system to treat glaucoma according to the visual field test. OCT investigates the structure of the optic nerve fiber layer in the eyes by non-invasive measurement, and the test takes only few minutes. In addition to the convenience, it detects the abnormality of structure changes which is more sensitive to early pathological changes than detecting the functional abnormality. Myopia, axial elongation of the eyeball, is associated with the appearance of an optic nerve crescent, disc tilt, disc torsion, and thinning of the peripapillary RNFL in some sectors, in the absence of glaucomatous change. 8-10 Those characteristic findings, which are usually associated with glaucoma, are found in high myopic ey...

Evaluation of peripapillary retinal nerve fiber layer thickness in patients with primary open-angle glaucoma

“…Similarly, according to this study, the best parameter of the RNFL thickness group is AvgThic with a largest area under the ROC curve 0.803, cut-off value of 63.9, sensitivity of 67%, and specificity of 83.3%. Hsieh et al [19] reported that the largest area below the ROC curve were with: AvgThic, quadrant I and quadrant S. Singh et al [14] also confirmed that the surface of the ROC curve is the largest for AvgThic and quadrant S parameter (Area = 0.963, Area = 0.943), and a slightly smaller area in the case of quadrant I, but with high values of sensitivity of 89% and specificity of 81%. These results are almost the same as results in our study which has underlined the importance of parameter Avg Thic, S, Smax, I and Iavg for the earliest possible diagnosis of glaucoma.…”

The Diagnostic Value of Optical Coherence Tomography for Detecting Glaucoma in High Myopia

“…(1) AUROC of RNFL (Figure S21–S37, Supplemental Digital Content 7, http://links.lww.com/IJG/A745). We detected 13 studies,14,16,22–24,26–32,34 8 studies 14,22,25–30,32 and 4 studies 23,24,29,31 measuring AUROC of average thickness of RNFL, 4 sectors of RNFL, 12-clock-hour sectors of RNFL, respectively. Pooled AUROC ranged from 0.528 to 0.922 depending on area imaged.…”

“…(1) AUROC of RNFL (Figure S21-S37, Supplemental Digital Content 7, http://links.lww.com/IJG/A745). We detected 13 studies, 14,16,[22][23][24][26][27][28][29][30][31][32]34 8 studies 14,22,[25][26][27][28][29][30]32 and 4 studies 23,24,29,31 AUROC in hemisphere. Hence, we did not perform an analysis on it.…”

Focused Analysis of RNFL Decay in Glaucomatous Eyes using Circular Statistics on High-resolution OCT data

“…While past studies have used spatial, trigonometric and Fourier analysis on OCT data, e.g., (22)(23)(24)(25), our framework, involving directional or circular data, is methodologically quite different from those approaches. While our earlier platform CIFU (14) focuses on the aspects of shapes and curves in OCT NRR data, as in functional data analysis (FDA), it does not explicitly address the directional characteristics of OCT data or angular measurements thereof.…”