Nukada, Masayuki MD; Hangai, Masanori MD; Mori, Satoshi MD; Takayama, Kohei MD; Nakano, Noriko MD; Morooka, Satoshi MD; Ikeda, Hanako O. MD; Akagi, Tadamichi MD; Nonaka, Atsushi MD; Yoshimura, Nagahisa MD
With glaucomatous optic neuropathy, there is progressive damage to the retinal ganglion cell axons within the optic nerve head, which diminishes the retinal nerve fiber layer (RNFL). RNFL defects can be seen before optic nerve head and visual-field damage become apparent.1–9 In fact, in various studies, RNFL defects were detected in more than half of the eyes with glaucoma about 6 years before any changes in the visual field were detected.2,5,7,8 Thus, a finding of early RNFL damage in eyes with preperimetric glaucoma may be predictive of visual-field loss.10
On high-resolution cross-sectional optical coherence tomography (OCT) fundus images, abnormal thinning of the circumpapillary RNFL (cpRNFL) has been demonstrated to be highly diagnostic of glaucoma.11–15 The most advanced OCT technology, spectral-domain OCT (SD-OCT), provides for faster imaging and better axial resolution compared with time-domain (TD, Stratus) OCT and therefore provides less variable and more highly reproducible measurements of cpRNFL.16–24 However, SD-OCT has not to date been proven to have better glaucoma diagnostic utility than TD-OCT, particularly in patients with early (preperimetric) glaucoma.25 It remains uncertain why SD-OCT has not led to improved detection of early RNFL damage.
In eyes with glaucoma, RNFL damage often can be seen as localized RNFL defects on biomicroscopic examination. In previous studies that sought to detect these RNFL defects by OCT, Jeoung et al26 found that TD-OCT has poor sensitivity for detection of localized RNFL defects with angular widths <10 degrees in eyes with glaucoma, and Kim et al27 found similar results for identifying cpRNFL defects by TD-OCT in eyes with preperimetric glaucoma. In another study, Jeoung and Park28 found that the sensitivities of SD-OCT (Cirrus) and TD-OCT (Stratus) were closely related to the angular widths of the preperimetric RNFL defects. Thus, the small angular width of preperimetric RNFL defects appears to be responsible for the poor utility of OCT to detect preperimetric RNFL defects.
In our previous study of eyes with perimetric glaucoma,29 we found that speckle noise, which blurs the boundaries between retinal layers,30–33 was responsible for unclear visualization of RNFL defects on tomographic images and for inaccurate delineation of the boundary between the RNFL and ganglion cell layer. Thus, speckle noise–reduced imaging techniques provide a unique opportunity to reveal characteristics of preperimetric RNFL defects in their tomographic images, which is possibly useful to improve the detection of preperimetric RNFL defects.
To determine whether SD-OCT instruments can reliably detect such early RNFL damage, we conducted the study reported here in which we analyzed speckle noise–reduced (multiple-scan averaged) SD-OCT B-scans of the cpRNFL obtained cross-sectionally in the areas where we had identified preperimetric RNFL defects on red-free fundus photographs. We compared these speckle noise–reduced images with TD-OCT and single-scan SD-OCT images to estimate the sensitivity of the 3 methods for detecting preperimetric RNFL defects and the sensitivity and specificity of the imaging methods for diagnosing glaucoma.
MATERIALS AND METHODS
Eyes with localized RNFL defects that met the eligibility criteria for this study were consecutively enrolled from the database of patients who were examined for glaucoma between January 2008 and December 2009 at the Department of Ophthalmology in Kyoto University Hospital. Control eyes for this study were those of healthy subjects who were determined by our department to have at least 1 normal eye and who agreed to undergo the examinations described in this study.
All patients and volunteers underwent a comprehensive ophthalmic examination, including measurement of uncorrected and best-corrected visual acuity using the 5 m Landolt chart, slit-lamp examinations, intraocular pressure (IOP) measurements using a Goldman applanation tonometer, gonioscopy, dilated biomicroscopic examination, stereo color disc photography (with a 3-Dx simultaneous stereo disc camera; Nidek, Gamagori, Japan), red-free fundus photography using the Heidelberg Retina Angiogram 2 (Heidelberg Engineering, Heidelberg, Germany), and standard automated perimetry (SAP) using the Humphrey Visual Field Analyzer (Carl Zeiss-Meditec, Dublin, CA).
Images analyzed for cpRNFL thickness were obtained using 3 types of OCT equipment: Stratus (time-domain; TD) OCT (Carl Zeiss-Meditec), RTvue-100 (single-scan SD-OCT) (Optovue, Fremont, CA), and Spectralis HRA+OCT (speckle noise–reduced by multiple-scan averaging SD-OCT with simultaneous retinal angiography) (Heidelberg Engineering).
To be included in this study, the eyes had to have a best-corrected visual acuity of 20/20 or better in Snellen equivalent or a spherical equivalent refractive error between −6 and 3 D; a normal open anterior chamber angle; good-quality red-free fundus photographs; acceptable signal strength of all the 3 OCT instruments (Stratus, >6; RTVue, >35, and Spectralis, >35), and reliable visual-field test results.
Patients with preperimetric glaucoma were included if they had normal SAP results and 1 or more localized wedge-shaped RNFL defects on red-free fundus photography that were associated with glaucomatous appearance of the optic nerve head (defined as diffuse or localized neuroretinal rim thinning) on stereo color fundus photographs.
To be included in the control group, volunteer eyes had to have IOP ≤21 mm Hg, no history of increased IOP, absence of glaucomatous disc appearance and no visible RNFL defect on red-free fundus photography, and normal SAP results.
Excluded from both groups were subjects with diabetes mellitus or any other systemic disease that might affect the eye as well as any eyes with a history of ocular surgery, evidence of vitreoretinal disease, uveitis, or nonglaucomatous optic neuropathy.
When both of a patient’s or volunteer’s eyes were eligible for this study, 1 eye was selected at random to be included.
All investigations in this study adhered to the tenets of the Declaration of Helsinki. This study was approved by the Institutional Review Board and Ethics Committee of the Kyoto University Graduate School of Medicine. Informed consent was obtained from the subjects after explanation of the nature and possible consequences of the study.
We defined reliable visual-field test results on SAP as fixation loss ≤20%, false positive ≤15%, and false negative ≤33%. Visual-field defects were defined using the 24-2 Swedish Interactive Threshold Algorithm standard program as (1) abnormal range on the glaucoma hemifield test; and (2) 3 abnormal points with P<5% probability of being normal, 1 with P<1% by pattern deviation or (3) pattern standard deviation of <5% of the normal reference, confirmed on 2 consecutive tests.
Optic Disc Evaluation
Three glaucoma specialists (M.H., H.O.I., and A.N.) who were masked to all other information about the eyes independently evaluated the appearance of the optic nerve head on fundus photographs, including stereoscopic photographs. Images of each eye from both normal and preperimetric glaucoma candidates were displayed on a monitor for evaluation. Specialists shared their independent decisions, and if the findings of all 3 were not in agreement, the group reached consensus by review and discussion of the color fundus photographs.
Measurement of Angular Width of RNFL Defects on Red-free Fundus Photographs
We measured the width of each RNFL defect identified on red-free fundus photographs by a technique we described previously.29 First, we drew on each red-free image a circle with a diameter of 3.46 mm centered on the optic nerve head. Next, we drew a line from the center of the optic nerve head to each point where the circle intersected the border of an RNFL defect. We obtained the angular width of the RNFL defect by measuring the angle between the pair of lines on either side of the RNFL defect.
Cross-Sectional Imaging of cpRNFL With Speckle Noise–reduced SD-OCT
We obtained cross-sectional speckle noise–reduced SD-OCT images of the cpRNFL through the area where RNFL defect(s) were identified on the red-free fundus photographs by performing repetitive B-scans in a circle with a diameter of 3.46 mm centered on the optic disc using the Spectralis HRA+OCT (software version 4.1). This instrument reduces speckle noise by simultaneously acquiring multiple SD-OCT (870 nm) B-scans and confocal laser scanning ophthalmoscope (cSLO) images with real-time 3-dimensional tracking of eye movements and real-time averaging of the B-scans acquired at an identical location of interest on the retina. Each circular B-scan for this study consisted of 1536 A-scans (acquired at a rate of 40,000/s), providing a digital transverse sampling resolution of 5 μm per pixel. For cpRNFL analysis, 16 scans were averaged to produce each image.29
Structural changes in the cpRNFL that corresponded to areas with RNFL defects seen on red-free fundus photography were identified by comparing red-free and speckle noise–reduced SD-OCT images.
Assessment of Accuracy for Automated Drawing of Boundary Lines
Two glaucoma specialists (K.T. and S.M.) who were masked to clinical information about the eyes determined whether the inner and outer boundary lines of the cpRNFL automatically drawn by the OCT software were correct in the area that corresponded to the localized RNFL defects on red-free fundus photographs. The interrater reliability was calculated for determinations by the 2 independent evaluators as to whether the software had drawn the boundary lines correctly or incorrectly. When the evaluations of the 2 specialists did not agree, a third glaucoma specialist (M.N.) examined the images, and the results were discussed by the 3 examiners until the group reached consensus.
cpRNFL Analysis Protocol
We analyzed color-coded maps and graphs that were automatically generated to display cpRNFL thickness by the applicable software program of each of the 3 OCT instruments: TD-OCT (Stratus OCT); standard, single-scan SD-OCT (RTVue-100); and speckle noise–reduced (by multiple-scan averaging) SD-OCT (Spectralis HRA+OCT).
We used the built-in software program of each OCT system to automatically measure cpRNFL thickness and calculate mean thickness over a defined area (sector) of the cpRNFL. For standard (single-scan) cpRNFL analysis on TD-OCT (Stratus), we used the fast cpRNFL program (256 A-scans×3 B-scans) to scan a circle with a diameter of 3.43 mm, centered on the optic nerve head; the Stratus software calculated the mean cpRNFL thickness in each of the 12 clock-hour sectors. For standard (single-scan) cpRNFL analysis on SD-OCT (RTVue-100, software v.4.0), we used the RNFL3.45 scan program (1024 A-scans×4 B-scans), and the software calculated the mean RNFL thickness in 16 circumferential (circumpapillary) sectors. For speckle noise–reduced cpRNFL analysis on SD-OCT (Spectralis, software v.4.0C), we used the cpRNFL circle scan program set to a diameter of 3.46 mm (1536 A-scans×1 B-scan) to obtain a mean RNFL thickness in each of the 6 circumferential (circumpapillary) sectors.
Each OCT software program creates 2 color-coded displays of cpRNFL thickness by comparing the mean measured thickness value with the mean in an age-matched normative database. On the sector map, a sector (each of the 12 sectors for Stratus, each of 16 sectors for RTVue, and each of 6 sectors for Spectralis) is displayed as green if the mean cpRNFL thickness in that sector is within the 95% confidence interval (CI) for normal eyes for that individual’s age. The sector is displayed as yellow (borderline) if the mean thickness is between the lower 95% CI and lower 99% CI (P<5% and ≥1%) for normal, and it is displayed as red (abnormal) if the mean thickness is below the lower 99% CI (P<1%) for normal eyes of that age.
The second color-coded display is the temporal-superior-nasal-inferior-temporal (TSNIT) graph of cpRNFL thickness. On this graph, the mean thickness measurements obtained circumferentially, moving from the temporal to the superior to the nasal to the inferior and back to the temporal sectors, are represented by a black line that is superimposed on bands of color that indicate 95% CI for normal cpRNFL thickness (green), <95% but>99% CI (P<5% and >1%) for borderline thickness (yellow), and abnormal thinning (<99% CI, P<1%) (red). We classified TSNIT graphs as showing abnormal (<1%) thinning of the cpRNFL when the cpRNFL thickness line touched the red band or was located below the yellow band. We categorized TSNIT graphs as showing borderline (<5%) cpRNFL thinning when the cpRNFL thickness line touched the yellow band or was located below the green band.
Sensitivity and Specificity
To compare sensitivity for detecting RNFL defects, we defined a true-positive result for “RNFL defect detection” as the presence of a red sector on the sector map or abnormal thinning of the cpRNFL on the TSNIT graph, in the location where an RNFL defect had been identified on red-free fundus photographs.29
To compare sensitivity and specificity of the OCT systems for glaucoma diagnosis, we defined a positive result for “glaucoma detected” as having at least 1 red sector on the sector map or abnormal thinning of the cpRNFL on the TSNIT graph.
All statistical analyses were performed using SPSS version 11.01 J. The statistical significance of differences between values for glaucoma versus control eyes was evaluated with an unpaired t test and a χ2 test. The χ2 test was also used to study the relationship between the angular width of preperimetric RNFL defects and the number of RNFL defects detected as abnormal cpRNFL thinning on sector maps or TSNIT graphs.
We used the Fisher exact test to compare the sensitivities and specificities of sector maps with those of the TSNIT graphs generated by the 3 OCT systems. The McNemar test was used to compare the sensitivities and specificities of the 3 OCT systems for RNFL defect detection and glaucoma detection. The level of statistical significance was set at P<0.05.
During the study period, 312 patients underwent OCT examinations (Stratus OCT, RTvue-100, and Spectralis) and SAP24-2 tests within a 2-month period at our clinic, and 45 eyes of 32 patients (17 men and 15 women; age range, 29 to 77 y, mean±SD=55.7±11.9 y) met criteria for inclusion in this study. After 1 eye was selected at random in patients with both eyes eligible for inclusion, for the 32 patient eyes in the study, the refractive error ranged from 0.75 to −6 D (mean±SD=−1.6±2.4 D) and visual-field mean deviation ranged from 2.36 to −3.33 dB (mean±SD=−0.15±1.53 dB). Of the 32 eyes, 24 had a single RNFL defect and 8 had 2 localized RNFL defects identified on red-free photographs, for a total of 40 RNFL defects in the 32 eyes.
The 30 normal subjects (17 men and 13 women; age range, 38 to 74 y, mean±SD=58.5±12.3 y) in the study were not statistically significantly different in sex distribution or age from the patients in the study. Their eyes were not significantly different in refractive error (range, 0.5 to −5.75 D; mean±SD=−1.2±2.3 D) or visual-field mean deviation (range, 1.39 to −1.88 dB; mean±SD=−0.11±0.95 dB) from the patient eyes.
Angular Width of Preperimetric RNFL Defects
All 40 of the RNFL defects identified on red-free fundus photographs had angular widths <30 degrees: 27.5% of RNFL defects were 10 degrees or less in angular width, and 70% were 20 degrees or less in angular width (Table 1).
Characteristics in Cross-sectional Images of Preperimetric RNFL Defects
On speckle noise–reduced images of the cpRNFL obtained using Spectralis, areas of diminished RNFL thickness were clearly visible in areas with preperimetric RNFL defects (Figs. 1, 2). The preperimetric RNFL defects varied in appearance: in some eyes, cpRNFL thinning appeared severe and abrupt (Fig. 1, first to fifth rows), whereas in others it appeared slight (Fig. 1, fifth to 10th rows). However, in no case did we see complete disruption of the band of reflectivity representing the RNFL.
Slight damage to the cpRNFL was less clearly seen in single-scan SD-OCT images (Fig. 2, second, fifth, eighth, and 11th rows) or single-scan TD-OCT images (Fig. 2, third, sixth, ninth, and bottom rows).
Automated Boundary-line Delineation of Preperimetric RNFL Thinning on OCT
In Figures 2 and 3, the OCT images show the boundary lines of the cpRNFL that were drawn on the images automatically by the software: lines are red on Spectralis (speckle noise–reduced SD-OCT) images and white on RTVue (standard, single-scan SD-OCT) and Stratus (single-scan TD-OCT) images.
Visual analysis of the boundary lines by 2 glaucoma specialists identified that the boundaries of the cpRNFL were correctly drawn by the software for 38 (95.0%) of 40 preperimetric RNFL defects seen on Spectralis SD-OCT images but for only 25 (62.5%) and 23 (57.5%) of the defects in images obtained by RTVue-100 SD-OCT or Stratus TD-OCT. The Cohen κ index of intercoder reliability was 0.802 (outstanding), 0.632 (substantial), and 0.689 (substantial) for agreement between the 2 specialists in boundary-line accuracy on Spectralis, RTVue-100, and TD-OCT images, respectively.
Effect of Accurate Boundary-line Delineation on RNFL Defect Detection
As shown in Figure 3, an RNFL defect in the superior sector on fundus photographs (top left) was not detected as thinning of the cpRNFL on sector maps generated by any of the 3 OCT imaging systems (bottom left: the 3 image maps have no yellow or red sectors).
On TSNIT thickness graphs, however, the cpRNFL thickness line dipped below the yellow band (indicating abnormal thinning) on Spectralis imaging, although the thickness line was within the 95% CI for normal on TSNIT graphs generated by single-scan SD-OCT and TD-OCT systems, representing false-negative results.
Effect of RNFL Angular Width on Sensitivity of Sector Maps and TSNIT Graphs
As shown in Table 1, in all OCT systems, detection of RNFL defects was worse when the angular width of the RNFL defect was smaller, but these differences according to angular width were not statistically significant (Table 1; P=0.244 to 0.501). Detection of RNFL defects was also worse on TSNIT thickness graphs with smaller angular widths, but the difference was not statistically significant (P=0.071 to 0.756).
OCT Type and Sensitivity of Sector Maps Versus TSNIT Graphs for Detecting Preperimetric RNFL Defects
As shown in Table 2, abnormal (P<1%) and borderline (P<5%) thinning of the cpRNFL were significantly more often matched to preperimetric RNFL defects on red-free fundus photographs when thinning was identified by SD-OCT (RTVue-100 or Spectralis) compared with TD-OCT (Stratus). The 2 SD-OCT imaging systems (RTVue and Spectralis) were equivalent in their sensitivity for detecting abnormal and borderline thinning of the cpRNFL.
TSNIT graphs were more sensitive than sector maps for detecting abnormal cpRNFL thinning on RTVue-100 and borderline and abnormal cpRNFL thinning on Spectralis but not on Stratus images (Table 2).
Analysis of TSNIT thickness graphs showed that abnormal (P<1%) and borderline (P<5%) thinning of the cpRNFL were significantly more often matched to preperimetric RNFL defects on red-free fundus photographs when thinning was identified by speckle noise–reduced SD-OCT (Spectralis) compared with single-scan SD-OCT (RTVue-100) or single-scan TD-OCT (Stratus), and when thinning was identified by single-scan SD-OCT (RTVue-100) compared with single-scan TD-OCT (Stratus).
OCT Type and Sensitivity/Specificity of Sector Maps Versus TSNIT Graphs for Glaucoma Diagnosis
As shown in Table 3, TSNIT thickness graphs were more sensitive for detecting glaucoma (<1%) than were sector maps for all OCT instruments; only in Spectralis, TSNIT thickness graphs were significantly more sensitive at the P<1% level for detecting glaucoma than were sector maps. TSNIT thickness graphs did not significantly reduce specificity, except for borderline cpRNFL thinning (P<5%) on Spectralis graphs.
Both the SD-OCT systems (RTVue-100 and Spectralis) were more sensitive than TD-OCT (Stratus) for glaucoma diagnosis based on sector maps (Table 3). There were no significant differences in sensitivity and specificity among the 3 OCT instruments when abnormal cpRNFL thinning on TSNIT thickness graph analysis was the criterion for glaucoma diagnosis.
In this report, we present characteristic speckle noise–reduced SD-OCT images that show cpRNFL thinning in locations corresponding to localized RNFL defects detected on fundus photographs of eyes with preperimetric glaucoma. Previously, we reported that some eyes with perimetric glaucoma had disruption of RNFL reflectivity in areas corresponding to localized RNFL defects,29 but in the study we report here, none of the eyes with preperimetric glaucoma evidenced disruption of RNFL reflectivity. Taken together with the fact that, in the current and previous studies, preperimetric RNFL defects were small in angular width, we speculate that morphologic characteristics of preperimetric RNFL defects may account for at least some cases of false-negative detection of abnormal cpRNFL thinning in preperimetric glaucoma.27,28
In our previous study of eyes with perimetric glaucoma, we found that OCT-generated boundary lines of the RNFL in areas where damage within RNFL defects did not disrupt RNFL reflectivity was inaccurate on single-scan SD-OCT and TD-OCT images but correct on speckle noise–reduced SD-OCT images,29 and that sector maps of single-scan OCT images had low sensitivity for detection of abnormal (<1%) thinning of the cpRNFL. In this previous study, speckle noise–reduced SD-OCT imaging was more sensitive than single-scan SD-OCT or TD-OCT in detecting abnormal thinning of the cpRNFL in areas with RNFL defects. Results in the current study were similar: Spectralis detected abnormal cpRNFL thinning in 47.5% of cases compared with 15.0% for TD-OCT and 42.5% for single-scan SD-OCT (Table 2). Nevertheless, the 47.5% sensitivity of Spectralis SD-OCT in this study was unexpectedly low; this could reflect greater difficulty detecting preperimetric RNFL defects compared with perimetric RNFL defects. These results are also consistent with the results of previous studies using Stratus OCT27 and Cirrus HD-OCT.28 Analysis by glaucoma experts of the software-generated outer boundary lines of the cpRNFL in this study found that it was accurate in 95.0% of speckle noise–reduced SD-OCT images [compared with only 62.5% of TD-OCT images and 57.5% of single-scan (RTVue) SD-OCT images]; therefore the low sensitivity of Spectralis SD-OCT in the current study is not attributable to inaccuracy in drawing boundary lines. On the basis of these findings, we suggest that the low sensitivity of Spectralis SD-OCT in detecting preperimetric RNFL defects may be due to morphologic characteristics of the defects we found, such as no disruption of their RNFL reflectivity and their small angular width.
In our previous study,29 42.8% of the perimetric RNFL defects in which the band of RNFL reflectivity was not disrupted on OCT images were >30 degrees (equivalent to 1 clock sector) in angular width, whereas in this study, all 40 of the preperimetric RNFL defects were <30 degrees in angular width: 28 (70%) of the 40 were 20 degrees or less, and 11 (27.5%) of the 40 were 10 degrees or less (Table 1). The widths of the preperimetric RNFL defects in our study are similar to those reported by Kim et al27 (81.0% to 94.1% were <30 degrees in angular width) and Jeoung and Park28 (80.6% were <20 degrees, and 29.0% were 10 degrees or less), and in these previous studies, as in ours, RNFL defects with smaller angular widths were much less likely to be detected on sector maps.27,28 This high false-negative detection rate for narrow preperimetric RNFL defects may result from the wide angular width of each sector defined by Spectralis (45 or 95 degrees), Cirrus (30 degrees), and Stratus (30 degrees) compared with the small angular width (≤10 or 20 degrees) of preperimetric RNFL defects. This effect would be even greater for RNFL defects that are partly in 1 sector and partly in an adjacent sector. In addition, slight cpRNFL thinning would make it more difficult to detect narrow RNFL defects with sector maps. We attribute the inability to detect preperimetric RNFL defects, particularly those <10 degrees in angle, to sector maps to both slight RNFL thinning and smaller defect width compared with the widths of the sectors.
TSNIT graphs of cpRNFL thickness avoid the problems just described for sector maps, because they show a continuous line representing cpRNFL thickness28; therefore it is not surprising that in our study, TSNIT graphs were more sensitive than sector maps for detection of abnormal and borderline cpRNFL thinning, although the difference was statistically significant only in Spectralis. TSNIT graphs were also significantly more sensitive without significant reduction in specificity for the detection of glaucoma (P<1%) by speckle noise–reduced SD-OCT (Table 3). Figure 3 shows an example of a preperimetric RNFL defect (<20 degrees in angular width) that was associated with slight thinning of the band of cpRNFL reflectivity on SD-OCT and that was not identified on the sector map but was identified on the TSNIT graph generated by speckle noise–reduced SD-OCT, although not on standard (single-scan) SD-OCT or TD-OCT.
A limitation of this study is that we excluded eyes with diffuse preperimetric RNFL atrophy, because this condition cannot be clearly identified with red-free photography. Thus, we do not know whether our findings regarding specificity and sensitivity of the 3 types of OCT would hold true for detecting RNFL changes in these eyes.
In conclusion, our study showed that both accurate measurement of cpRNFL thickness by speckle noise–reduced SD-OCT and comparison of the results with age-matched normative database using TSNIT graphs together improve the sensitivity of SD-OCT for detecting narrow preperimetric RNFL defects and preperimetric glaucoma diagnosis. It was shown that analysis of the RNFL thickness deviation map on cube scan (3-dimensional raster scan based on single-scan images) would be more sensitive for glaucoma detection compared with conventional cpRNFL analysis.28,34 However, imaging speed of currently available SD-OCT instruments does not allow 3-dimensional raster scanning based on speckle noise–reduced images. Further enhancements to the use of OCT to diagnose preperimetric glaucoma most likely need to await the development of devices with imaging speeds fast enough to allow combined speckle noise reduction and cube scanning.
1. Hoyt WF, Newman NM.The earliest observable defect in glaucoma?Lancet.1972;1:692–693.
2. Sommer A, Miller NR, Pollack I, et al..The nerve fiber layer in the diagnosis of glaucoma.Arch Ophthalmol.1977;95:2149–2156.
3. Sommer A, Pollack I, Maumenee AE.Optic disc parameters and onset of glaucomatous field loss. I. Methods and progressive changes in disc morphology.Arch Ophthalmol.1979;97:1444–1448.
4. Funk J.Early detection of glaucoma by longitudinal monitoring of the optic disc structure.Graefes Arch Clin Exp Ophthalmol.1991;229:57–61.
5. Sommer A, Katz J, Quigley HA, et al..Clinically detectable nerve fiber atrophy precedes the onset of glaucomatous field loss.Arch Ophthalmol.1991;109:77–83.
6. Motolko M, Drance SM.Features of the optic disc in preglaucomatous eyes.Arch Ophthalmol.1981;99:1992–1994.
7. Tuulonen A, Airaksinen PJ.Initial glaucomatous optic disk and retinal nerve fiber layer abnormalities and their progression.Am J Ophthalmol.1991;111:485–490.
8. Tuulonen A, Lehtola J, Airaksinen PJ.Nerve fiber layer defects with normal visual fields. Do normal optic disc and normal visual field indicate absence of glaucomatous abnormality?Ophthalmology.1993;100:587–597.
9. Kass MA, Heuer DK, Higginbotham EJ, et al..The Ocular Hypertension Treatment Study: A randomized trial determines that topical ocular hypotensive medication delays or prevents the onset of primary open-angle glaucoma.Arch Ophthalmol.2002;120:701–713.
10. Quigley HA, Miller NR, George T.Clinical evaluation of nerve fiber layer atrophy as an indicator of glaucomatous optic nerve damage.Arch Ophthalmol.1980;98:1564–1571.
11. Guedes V, Schuman JS, Hertzmark E, et al..Optical coherence tomography measurement of macular and nerve fiber layer thickness in normal and glaucomatous human eyes.Ophthalmology.2003;110:177–189.
12. Wollstein G, Schuman JS, Price LL, et al..Optical coherence tomography (OCT) macular and peripapillary retinal nerve fiber layer measurements and automated visual fields.Am J Ophthalmol.2004;138:218–225.
13. Wollstein G, Ishikawa H, Wang J, et al..Comparison of three optical coherence tomography scanning areas for detection of glaucomatous damage.Am J Ophthalmol.2005;139:39–43.
14. Medeiros FA, Zangwill LM, Bowd C, 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;139:44–55.
15. Ojima T, Tanabe T, Hangai M, et al..Measurement of retinal nerve fiber layer thickness and macular volume for glaucoma detection using optical coherence tomography.Jpn J Ophthalmol.2007;51:197–203.
16. González-García AO, Vizzeri G, Bowd C, et al..Reproducibility of RTVue retinal nerve fiber layer thickness and optic disc measurements and agreement with Stratus optical coherence tomography measurements.Am J Ophthalmol.2009;147:1067–10741074.e1.
17. Hong S, Kim CY, Lee WS, et al..Reproducibility of peripapillary retinal nerve fiber layer thickness with spectral domain Cirrus high-definition optical coherence tomography in normal eyes.Jpn J Ophthalmol.2010;54:43–47.
18. Leung CK, Cheung CY, Weinreb RN, et al..Retinal nerve fiber layer imaging with spectral-domain optical coherence tomography: a variability and diagnostic performance study.Ophthalmology.2009;116:1257–12631263.e1–1263.e2.
19. Lee SH, Kim SH, Kim TW, et al..Reproducibility of retinal nerve fiber thickness measurements using the test-retest function of spectral OCT/SLO in normal and glaucomatous eyes.J Glaucoma.2010;19:637–642.
20. Garas A, Vargha P, Holló G.Reproducibility of retinal nerve fiber layer and macular thickness measurement with the RTVue-100 optical coherence tomograph.Ophthalmology.2010;117:738–746.
21. Mwanza JC, Chang RT, Budenz DL, et al..Reproducibility of peripapillary retinal nerve fiber layer thickness and optic nerve head parameters measured with cirrus HD-OCT in glaucomatous eyes.Invest Ophthalmol Vis Sci.2010;51:5724–5730.
22. Nakatani Y, Higashide T, Ohkubo S, et al..Evaluation of macular thickness and peripapillary retinal nerve fiber layer thickness for detection of early glaucoma using spectral domain optical coherence tomography.J Glaucoma.2011;20:252–259.
23. Wu H, de Boer JF, Chen TC.Reproducibility of retinal nerve fiber layer thickness measurements using spectral domain optical coherence tomography.J Glaucoma.2011;20:470–476.
24. Tan BB, Natividad M, Chua KC, et al..Comparison of retinal nerve fiber layer measurement between 2 spectral domain OCT instruments.J Glaucoma.2012;21:266–273.
25. Savini G, Carbonelli M, Barboni P.Spectral-domain optical coherence tomography for the diagnosis and follow-up of glaucoma.Curr Opin Ophthalmol.2011;22:115–123.
26. Jeoung JW, Park KH, Kim TW, et al..Diagnostic ability of optical coherence tomography with a normative database to detect localized retinal nerve fiber layer defects.Ophthalmology.2005;112:2157–2163.
27. Kim TW, Park UC, Park KH, et al..Ability of Stratus OCT to identify localized retinal nerve fiber layer defects in patients with normal standard automated perimetry results.Invest Ophthalmol Vis Sci.2007;48:1635–1641.
28. Jeoung JW, Park KH.Comparison of Cirrus OCT and Stratus OCT on the ability to detect localized retinal nerve fiber layer defects in preperimetric glaucoma.Invest Ophthalmol Vis Sci.2010;51:938–945.
29. Nukada M, Hangai M, Mori S, et al..Detection of localized retinal nerve fiber layer defects in glaucoma using enhanced spectral-domain optical coherence tomography.Ophthalmology.2011;118:1038–1048.
30. Schmitt JM, Xiang SH, Yung KM.Speckle in optical coherence tomography.J Biomed Optics.1999;4:95–105.
31. Sander B, Larsen M, Thrane L, et al..Enhanced optical coherence tomography imaging by multiple scan averaging.Br J Ophthalmol.2005;89:207–212.
32. Sakamoto A, Hangai M, Yoshimura N.Spectral-domain optical coherence tomography with multiple B-scan averaging for enhanced imaging of retinal diseases.Ophthalmology.2008;115:1071–1078.
33. Hangai M, Yamamoto M, Sakamoto A, et al..Ultrahigh-resolution versus speckle noise-reduction in spectral-domain optical coherence tomography.Opt Express.2009;17:4221–4235.
34. Leung CK, Lam S, Weinreb RN, et al..Retinal nerve fiber layer imaging with spectral-domain optical coherence tomography: analysis of the retinal nerve fiber layer map for glaucoma detection.Ophthalmology.2010;117:1684–1691.
© 2014 by Lippincott Williams & Wilkins.