Diagnostic Classification of Retinal Nerve Fiber Layer Measurement in Myopic Eyes: A Comparison Between Time-Domain and Spectral-Domain Optical Coherence Tomography

doi:10.1016/j.ajo.2011.04.002

American Journal of Ophthalmology

Volume 152, Issue 4, October 2011, Pages 646-653.e2

https://doi.org/10.1016/j.ajo.2011.04.002 Get rights and content

Purpose

To evaluate and compare the diagnostic classification of retinal nerve fiber layer (RNFL) measurement between time-domain and spectral-domain optical coherence tomography (OCT) in myopic eyes.

Design

Prospective, observational study.

Methods

A total of 97 eyes from 97 healthy myopic subjects were included. The RNFL in each eye was imaged sequentially with the Stratus OCT and the Cirrus HD-OCT (Carl Zeiss Meditec). With reference to the built-in normative database, the number of abnormal diagnostic classifications (borderline or outside normal limits) in each OCT device was analyzed and compared using the likelihood ratio chi-square test. Multiple logistic regression analysis was performed to evaluate factors associated with abnormal diagnostic classification.

Results

The Cirrus HD-OCT classified a significantly higher percentage of eyes as outside normal limits/borderline in at least 1 clock hour (Stratus, 14.4%/24.8%; Cirrus, 21.6%/34.1%; all P < .01). RNFL measurement at 1 (23.6%) and 2 o'clock (23.5%) of all eyes was the most frequent location classified as abnormal by the Cirrus HD-OCT and the Stratus OCT, respectively. Eyes with smaller optic disc and longer axial length were more likely to have abnormal diagnostic classification.

Conclusions

In myopic eyes, Cirrus HD-OCT was more likely to have abnormal diagnostic classification than the Stratus OCT.

Section snippets

Subjects

One hundred Chinese myopic subjects were recruited in our study. One eye from each subject was randomly selected. All subjects received complete ophthalmic examinations in Joint Shantou International Eye Center, which included visual acuity, intraocular pressure (IOP) measurement, refraction, axial length measurement by IOL master (Carl Zeiss Meditec), and dilated fundus stereoscopic examination. All the included eyes have spherical equivalent of less than −0.5 diopters (D) and no other

Results

After excluding 2 subjects with unreliable visual field tests and 1 for extensive PPA, 97 eyes from 97 subjects (43 female and 61 right eyes) were analyzed. The mean age, axial length, and spherical equivalent were 22.84 ± 3.92 years (range, 18-40), 25.55 ± 1.11 mm (range, 22.52–28.77 mm), and −4.93 ± 2.17 D (range, −1.00 to −12.75 D), respectively. The visual field mean deviation and pattern standard deviation were −2.13 ± 1.12 dB and 1.49 ± 0.98 dB. The signal strength of Stratus OCT was

Discussion

In this study, we found that nasal RNFL measurement was frequently classified as borderline or outside normal limits in both OCTs and that the frequency of abnormal classification of RNFL measurement was significantly higher in Cirrus HD-OCT than Stratus OCT. Eyes with smaller optic disc size and longer axial length were found to be associated with significantly higher percentages of abnormal diagnostic classification in both devices.

The Stratus OCT is a time-domain OCT that has been shown

Mingzhi Zhang, is a Professor of Glaucoma at Joint Shantou International Eye Center of Shantou University and The Chinese University of Hong Kong. She is experienced in glaucoma surgery and cataract and refractive surgery. Dr Zhang's primary research focus is ocular imaging and neuroprotection in glaucoma.

References (30)

P. Mitchell et al.
The relationship between glaucoma and myopia: the Blue Mountains Eye Study
Ophthalmology
(1999)
Y.B. Liang et al.
Refractive errors in a rural chinese adult population: the Handan Eye Study
Ophthalmology
(2009)
F.A. Medeiros et al.
Evaluation of retinal nerve fiber layer, optic nerve head, and macular thickness measurements for glaucoma detection using optical coherence tomography
Am J Ophthalmol
(2005)
D.L. Budenz et al.
Sensitivity and specificity of the Stratus OCT for perimetric glaucoma
Ophthalmology
(2005)
C.K. Leung et al.
Retinal nerve fiber layer imaging with spectral-domain optical coherence tomography: a study on diagnostic agreement with Heidelberg Retinal Tomograph
Ophthalmology
(2010)
J.W. Jeoung et al.
Diagnostic ability of optical coherence tomography with a normative database to detect localized retinal nerve fiber layer defects
Ophthalmology
(2005)
S.T. Hoh et al.
Peripapillary retinal nerve fiber layer thickness variations with myopia
Ophthalmology
(2006)
F.S. Mikelberg et al.
Relation between optic nerve axon number and axon diameter to scleral canal area
Ophthalmology
(1991)
D.L. Budenz et al.
Determinants of normal retinal nerve fiber layer thickness measured by Stratus OCT
Ophthalmology
(2007)
J. Katz et al.
Prevalence and risk factors for refractive errors in an adult inner city population
Invest Ophthalmol Vis Sci
(1997)

A. Sommer et al.

Clinically detectable nerve fiber atrophy precedes the onset of glaucomatous field loss

Arch Ophthalmol

(1991)

D. Huang et al.

Optical coherence tomography

Science

(1991)

J.S. Schuman et al.

Quantification of nerve fiber layer thickness in normal and glaucomatous eyes using optical coherence tomography

Arch Ophthalmol

(1995)

F.A. Medeiros et al.

Comparison of the GDx VCC scanning laser polarimeter, HRT II confocal scanning laser ophthalmoscope, and Stratus OCT optical coherence tomograph for the detection of glaucoma

Arch Ophthalmol

(2004)

C.K. Leung et al.

Retinal nerve fiber layer measurements in myopia: An optical coherence tomography study

Invest Ophthalmol Vis Sci

(2006)

Cited by (27)

Detection of glaucomatous optic neuropathy with spectral-domain optical coherence tomography: a retrospective training and validation deep-learning analysis
2019, The Lancet Digital Health
Citation Excerpt :
Furthermore, diagnosis of glaucoma based on thickness of the retinal nerve fibre layer and built-in normative databases might not be reliable, because the layered thickness is affected by various factors including axial length, myopia, age, optic disc size, and signal strength.12–15 For example, eyes with small optic discs and long optic axes are associated with significantly increased percentages of abnormal diagnostic classification based on an SDOCT built-in normative database,16 resulting in false-positive errors. Deep learning with convolutional neural networks has been extensively used in medical imaging—particularly for pattern recognition and image classification.17
Spectral-domain optical coherence tomography (SDOCT) can be used to detect glaucomatous optic neuropathy, but human expertise in interpretation of SDOCT is limited. We aimed to develop and validate a three-dimensional (3D) deep-learning system using SDOCT volumes to detect glaucomatous optic neuropathy.
We retrospectively collected a dataset including 4877 SDOCT volumes of optic disc cube for training (60%), testing (20%), and primary validation (20%) from electronic medical and research records at the Chinese University of Hong Kong Eye Centre (Hong Kong, China) and the Hong Kong Eye Hospital (Hong Kong, China). Residual network was used to build the 3D deep-learning system. Three independent datasets (two from Hong Kong and one from Stanford, CA, USA), including 546, 267, and 1231 SDOCT volumes, respectively, were used for external validation of the deep-learning system. Volumes were labelled as having or not having glaucomatous optic neuropathy according to the criteria of retinal nerve fibre layer thinning on reliable SDOCT images with position-correlated visual field defect. Heatmaps were generated for qualitative assessments.
6921 SDOCT volumes from 1 384 200 two-dimensional cross-sectional scans were studied. The 3D deep-learning system had an area under the receiver operation characteristics curve (AUROC) of 0·969 (95% CI 0·960–0·976), sensitivity of 89% (95% CI 83–93), specificity of 96% (92–99), and accuracy of 91% (89–93) in the primary validation, outperforming a two-dimensional deep-learning system that was trained on en face fundus images (AUROC 0·921 [0·905–0·937]; p<0·0001). The 3D deep-learning system performed similarly in the external validation datasets, with AUROCs of 0·893–0·897, sensitivities of 78–90%, specificities of 79–86%, and accuracies of 80–86%. The heatmaps of glaucomatous optic neuropathy showed that the learned features by the 3D deep-learning system used for detection of glaucomatous optic neuropathy were similar to those used by clinicians.
The proposed 3D deep-learning system performed well in detection of glaucomatous optic neuropathy in both primary and external validations. Further prospective studies are needed to estimate the incremental cost-effectiveness of incorporation of an artificial intelligence-based model for glaucoma screening.
Hong Kong Research Grants Council.
Assessment of Optical Coherence Tomography Color Probability Codes in Myopic Glaucoma Eyes After Applying a Myopic Normative Database
2017, American Journal of Ophthalmology
Diagnostic classification of macular ganglion cell and retinal nerve fiber layer analysis: Differentiation of false-positives from glaucoma
2015, Ophthalmology
Citation Excerpt :
Despite the relatively small number of eyes (n = 39) with abnormal GCA deviation maps, we believe that these findings can provide valuable information to clinicians for differentiating between real and false-positive GCIPL abnormalities. The rate and associated factors of abnormal RNFL diagnostic classification have been investigated.6,8–10 Although the criteria of abnormal RNFL maps were different from ours, Yamashita et al8 and Kim et al16 reported that longer axial lengths and smaller disc areas were associated with false-positive RNFL maps, which is in accordance with our results.
To investigate the rate and associated factors of false-positive diagnostic classification of ganglion cell analysis (GCA) and retinal nerve fiber layer (RNFL) maps, and characteristic false-positive patterns on optical coherence tomography (OCT) deviation maps.
Prospective, cross-sectional study.
A total of 104 healthy eyes of 104 normal participants.
All participants underwent peripapillary and macular spectral-domain (Cirrus-HD, Carl Zeiss Meditec Inc, Dublin, CA) OCT scans. False-positive diagnostic classification was defined as yellow or red color-coded areas for GCA and RNFL maps. Univariate and multivariate logistic regression analyses were used to determine associated factors. Eyes with abnormal OCT deviation maps were categorized on the basis of the shape and location of abnormal color-coded area. Differences in clinical characteristics among the subgroups were compared.
(1) The rate and associated factors of false-positive OCT maps; (2) patterns of false-positive, color-coded areas on the GCA deviation map and associated clinical characteristics.
Of the 104 healthy eyes, 42 (40.4%) and 32 (30.8%) showed abnormal diagnostic classifications on any of the GCA and RNFL maps, respectively. Multivariate analysis revealed that false-positive GCA diagnostic classification was associated with longer axial length and larger fovea-disc angle, whereas longer axial length and smaller disc area were associated with abnormal RNFL maps. Eyes with abnormal GCA deviation map were categorized as group A (donut-shaped round area around the inner annulus), group B (island-like isolated area), and group C (diffuse, circular area with an irregular inner margin in either). The axial length showed a significant increasing trend from group A to C (P = 0.001), and likewise, the refractive error was more myopic in group C than in groups A (P = 0.015) and B (P = 0.014). Group C had thinner average ganglion cell–inner plexiform layer thickness compared with other groups (group A = B > C, P = 0.004).
Abnormal OCT diagnostic classification should be interpreted with caution, especially in eyes with long axial lengths, large fovea-disc angles, and small optic discs. Our findings suggest that the characteristic patterns of OCT deviation map can provide useful clues to distinguish glaucomatous changes from false-positive findings.
Diagnostic capability of peripapillary retinal thickness in glaucoma using 3D volume scans
2015, American Journal of Ophthalmology
Citation Excerpt :
In summary, compared to RNFL thickness, RT is consistently easier to measure accurately in glaucoma patients, especially in those with moderate to severe glaucoma. Peripapillary RT measurements in glaucoma may have some other possible advantages over RNFL measurements, especially if there are conditions that may decrease the accuracy of RNFL measurements, such as myopia, PPA, and severe glaucoma.15,16,18,21,41 In this study, we specifically looked at PPA, in order to determine if PPA affects the diagnostic capability of RT measurements.
To determine the diagnostic capability of spectral-domain optical coherence tomography (SD OCT) peripapillary retinal thickness (RT) measurements from 3-dimensional (3D) volume scans for primary open-angle glaucoma (POAG).
Cross-sectional study.
setting: Institutional. study population: 156 patients (89 POAG and 67 normal subjects). observation procedures: One eye of each subject was included. SD OCT peripapillary RT values from 3D volume scans were calculated for 4 quadrants of 3 different sized annuli. Peripapillary retinal nerve fiber layer (RNFL) thickness values were also determined. main outcome measures: Area under the receiver operating characteristic curve (AUROC) values, sensitivity, specificity, positive and negative predictive values, and positive and negative likelihood ratios.
The top 5 RT AUROCs for all glaucoma patients and for a subset of early glaucoma patients were for the inferior quadrant of outer circumpapillary annulus of circular grid (OCA) 1 (0.959, 0.939), inferior quadrant of OCA2 (0.945, 0.921), superior quadrant of OCA1 (0.890, 0.811), inferior quadrant of OCA3 (0.887, 0.854), and superior quadrant of OCA2 (0.879, 0.807). Smaller RT annuli OCA1 and OCA2 consistently showed better diagnostic performance than the larger RT annulus OCA3. For both RNFL and RT measurements, best AUROC values were found for inferior RT OCA1 and OCA2, followed by inferior and overall RNFL thickness.
Peripapillary RT measurements from 3D volume scans showed excellent diagnostic performance for detecting both glaucoma and early glaucoma patients. Peripapillary RT values have the same or better diagnostic capability compared to peripapillary RNFL thickness measurements, while also having fewer algorithm errors.
Macular imaging in highly myopic eyes with and without glaucoma
2013, American Journal of Ophthalmology
To determine how evaluations of macular structures on spectral-domain optical coherence tomography compare with those of the optic disc and circumpapillary retinal nerve fiber layer (RNFL) in discriminating between highly myopic eyes with and without glaucoma.
Retrospective, comparative study.
The appearances of ganglion cell layer and RNFL on Spectralis macular scans (Heidelberg Engineering) and optic disc on photographs were evaluated by 2 observers. The receiver operating characteristic regression was conducted for macular ganglion cell complex and circumpapillary RNFL measurements on RTVue-100 (Optovue).
Ninety highly myopic eyes (−6.0 to −15.0 diopters; mean deviation [MD], −5.6 ± 5.1 dB) and 91 non–highly myopic eyes (1.0 to −5.5 diopters; MD, −4.9 ± 5.7 dB) were enrolled. In highly myopic eyes (<−6 diopters), the Cohen κ for qualitative decisions by observers was 0.363 for photographs and 0.946 for Spectralis macular scans, and observers' evaluations of Spectralis macular scans were more accurate (94.5% and 94.5%, respectively; P < .0001) than their evaluations of photographs (71.4% and 80.2%, respectively). In the receiver operating characteristic regression analyses assessing the influences of age, sex, MD, and axial length, the better MD (P = .002 to .016) and longer axial length (P = .031 to .041) were associated significantly with diagnostic performances for all or some spectral-domain optical coherence tomography parameters. The receiver operating characteristic curves of average macular ganglion cell complex and circumpapillary RNFL thicknesses were comparable at low MD.
In high myopes, observers' assessments of the spectral-domain optical coherence tomography macular scans may agree better and be more accurate than observers' optic disc assessments. Glaucoma diagnostic performance of the macular ganglion cell complex may be less affected by axial length compared with that of circumpapillary RNFL.
From the past to the present, optical coherence tomography in glaucoma: a practical guide to a common disease
2024, F1000Research

View all citing articles on Scopus

Kunliang Qiu, is currently a resident at Department of Glaucoma, Joint Shantou International Eye Center of Shantou University and The Chinese University of Hong Kong. Dr Qiu is interested in ocular imaging with hopes of becoming a proliferative clinician-scientist.

View full text

Original articleDiagnostic Classification of Retinal Nerve Fiber Layer Measurement in Myopic Eyes: A Comparison Between Time-Domain and Spectral-Domain Optical Coherence Tomography

Purpose

Design

Methods

Results

Conclusions

Section snippets

Subjects

Results

Discussion

Ophthalmology

Ophthalmology

Am J Ophthalmol

Ophthalmology

Ophthalmology

Ophthalmology

Ophthalmology

Ophthalmology

Ophthalmology

Prevalence and risk factors for refractive errors in an adult inner city population

Invest Ophthalmol Vis Sci

Clinically detectable nerve fiber atrophy precedes the onset of glaucomatous field loss

Arch Ophthalmol

Optical coherence tomography

Science

Quantification of nerve fiber layer thickness in normal and glaucomatous eyes using optical coherence tomography

Arch Ophthalmol

Comparison of the GDx VCC scanning laser polarimeter, HRT II confocal scanning laser ophthalmoscope, and Stratus OCT optical coherence tomograph for the detection of glaucoma

Arch Ophthalmol

Retinal nerve fiber layer measurements in myopia: An optical coherence tomography study

Invest Ophthalmol Vis Sci

Original article
Diagnostic Classification of Retinal Nerve Fiber Layer Measurement in Myopic Eyes: A Comparison Between Time-Domain and Spectral-Domain Optical Coherence Tomography