Various optical coherence tomography (OCT) devices produced by different manufacturers are available to aid in clinical assessment of glaucomatous individuals, and a comparison of commonly used devices is discussed here. This chapter outlines differences in the imaging specifications, analysis techniques, normative databases, and diagnostic capabilities among the Cirrus 6000 (Carl Zeiss Meditec AG, Jena, Germany), Standard and Glaucoma Module Premium Editions of the Spectralis (Heidelberg Engineering GmbH, Heidelberg, Germany), Avanti RTVue XR (Optovue, Inc., Fremont, CA, USA), and 3D OCT (Topcon Corporation, Tokyo, Japan) imaging platforms. Differences are highlighted according to measurement and analysis of retinal nerve fiber layer, optic nerve head, and macular parameters.
Key wordsglaucoma – imaging – optical coherence tomography – retinal nerve fiber layer – optic nerve head – macula – Cirrus – Spectralis – Avanti – Topcon
7 Comparison of Common Devices
7.1 Retinal Nerve Fiber Layer Thickness
All of these optical coherence tomography (OCT) devices offer a scanning pattern that provides data on retinal nerve fiber layer (RNFL) thickness. The scan patterns offered by the devices can be divided into circle scans and volume scans. Circle scans measure circumpapillary RNFL (cpRNFL) thickness using a dedicated circular scanning pattern centered on the optic nerve head (ONH). Devices employing this strategy include the Spectralis (Heidelberg Engineering, Inc., Heidelberg, Germany), the Avanti RTVue XR (Optovue, Inc., Fremont, CA, USA), and the 3D OCT-2000 (Topcon Corp., Tokyo, Japan). In contrast, volume scans (also known as cube scans) secondarily derive a calculation circle to measure cpRNFL thickness. The volume scan, however, affords the benefit of providing a thickness map detailing the RNFL thickness across the entire scanned area. Devices employing this strategy include the Cirrus 6000 (Carl Zeiss Meditec AG, Jena, Germany), and the 3D OCT-2000 and OCT-1 Maestro2 (Topcon Corp., Tokyo, Japan). Devices may also be divided on the basis of their scanning and/or analytical axis, with some aligned to the horizontal axis and others to the fovea-disc (FoDi) axis. Although the chosen axis may affect RNFL thickness measurements, aligning analyses to the FoDi axis may not improve glaucoma detection. 1
Studies comparing the RNFL metrics of multiple devices have shown that, although they have been shown to be comparable, the diagnostic capabilities of devices may exhibit slight differences, including throughout a range of disease stages and myopia. 2 , 3 , 4 , 5 Diagnostic capabilities generally increase with more advanced stages of glaucoma and are least with preperimetric and mild glaucoma. (See subsequent subsections for details on each device.) 5 Note that literature cited in the following sections includes studies on single devices; as such, direct comparisons of study outcomes cannot be made across devices. Table 7‑1 highlights the main differences among the various platforms in imaging the RNFL and ONH.
7.1.1 Cirrus 6000 (Carl Zeiss Meditec AG, Jena, Germany)
The Cirrus 6000 is the latest OCT model from Carl Zeiss Meditec AG. Imaging specifications 6 include scanning speeds of 100,000 A-scans per second, an improvement over the prior model, the Cirrus 5000, which attains scanning speeds of 27–68,000 A-scans per second. For RNFL metrics, the device employs a 6.0 × 6.0 mm volume scan comprising 40,000 data points (200 horizontal B-scans each comprising 200 A-scans). The scan pattern is manually centered over the ONH. The software then derives a calculation circle with diameter estimated to be 3.46 mm and comprising 256 data points. 7 The calculation circle is automatically centered over the ONH by identifying the Bruch’s membrane opening (BMO) and then identifying the center of corresponding enclosed area.
The Cirrus 6000 offers a comprehensive analysis report, the ONH and RNFL Analysis 6 (Fig. 7‑1c and Fig. 7‑2c). The report displays an RNFL thickness map, where cooler colors (blue and green) represent thinner RNFL and warmer colors (red and yellow) represent thicker RNFL. Thickness colors range from 0 (blue) to 350 μm (white). Adjacent to the thickness map is displayed a deviation map, which compares regions of the thickness map to an age-matched normative database. Pixels are binned into “superpixels” of 16 pixels (4 × 4 pixels) for a total of 2,500 superpixels analyzed. 6
The deviation map is color-coded with respect to deviation from the normative database, whereby yellow signifies an RNFL thickness seen in the bottom 1 to 5% of normals, and red signifies an RNFL thickness seen in the bottom 1% of normals. Given that 5% of normals will have yellow- or red-range RNFL thickness, out of 2,500 superpixels analyzed, approximately 125 of these can be expected to be highlighted even in normals.
In addition to the RNFL thickness and deviation maps, cpRNFL thickness measured from the calculation circle is displayed in sector grids (both four- and twelve-sector) and a thickness graph (colloquially known as a “TSNIT” graph). The thickness graph plots cpRNFL thickness as a function of the position along the optic disc circumference, with 0 degree located at the horizontal axis. The cpRNFL thickness values are compared to an age-matched normative database, with the same yellow–red percentile deviation scheme as the deviation map. For review, the calculation circle tomogram is also displayed, as is a single B-scan of the volume scan.
In the data table, the average cpRNFL thickness (around the calculation circle) and the inter-eye asymmetry are displayed. The symmetry analysis is a correlation coefficient, converted to a percentage, that compares the 256 data points of the right eye calculation circle to the corresponding points of the left eye.
The Cirrus RNFL and Macula Normative Database 6 included males or females aged 18 years or older with a normal Humphrey 24–2 SITA Standard visual field in both eyes. Ocular exclusion criteria included: best-corrected visual acuity in either eye worse than 20/40; refractive error (spherical equivalent) outside of −12.00 D to +8.00 D; presence or history of intraocular pressure ≥22 mmHg in either eye; a diagnosis of glaucoma or glaucoma suspect in either eye; an occludable angle or a history of angle-closure in either eye; presence or history of disc hemorrhage in either eye; presence of RNFL defect in either eye; presence of amblyopia in either eye; previous laser or incisional surgery; any active infection of anterior or posterior segments; or evidence of diabetic retinopathy, diabetic macular edema, or other vitreoretinal disease. Systemic exclusion criteria included: history of diabetes, leukemia, AIDS, uncontrolled systemic hypertension, dementia, or multiple sclerosis; a life-threatening or debilitating disease; current or recent (within the past 14 days) use of an agent with photosensitizing properties by any route. Images were obtained of those with a signal strength of 5 or lower, large motion artifact, >10% area of data loss, and/or a floater obstructing appropriate view.
The Cirrus RNFL and Macula Normative Database 6 comprises 284 subjects: 134 males and 150 females, aged 19 to 84. Only 28 subjects were aged 70 to 79, and only 3 subjects were 80 years of age or older. Thus, results in patients aged 70 or above should be interpreted with caution. The normative database also exhibits the following ethnic breakdown: 43% Caucasian, 24% Asian, 18% African-American, 12% Hispanic, 1% Indian, and 2% mixed ethnicity.
Among all variables studied, age was found to have the greatest effect on the RNFL thickness metrics. For this reason, all results are corrected for age.
Numerous studies have evaluated the reliability and diagnostic capabilities of cpRNFL metrics on the Cirrus, generally on prior models with similar imaging specifications and analytics, albeit with slower scanning speeds. Overall, the area under the receiver operating characteristic curve (AUC, listed parenthetically) to distinguish all-stage glaucoma from normal reaches a maximum for average (0.964, 0.950, 0.921, 0.88), superior (0.933, 0.906, 0.891, 0.88), and inferior (0.953, 0.952, 0.924, 0.87) cpRNFL thickness. 2 , 3 , 7 , 8 AUCs to distinguish mild glaucoma from normal were greatest for inferior (0.923, 0.895, 0.857), average (0.940, 0.893, 0.856), and superior (0.872, 0.859, 0.842) cpRNFL thickness. 3 , 7 , 8 Considering the specific case of preperimetric glaucoma, the parameters with the best discriminating capability were average (0.752) and inferior (0.744) cpRNFL thickness. 9
Additional studies have evaluated the diagnostic capability of cpRNFL thickness in myopic eyes. The parameters with the highest discriminating capabilities reside in the average (0.899) and inferior (0.906) cpRNFL thickness in highly myopic glaucomatous eyes and the average cpRNFL thickness (0.920) in nonhighly myopic eyes. 10
7.1.2 Spectralis (Heidelberg Engineering GmbH, Heidelberg, Germany)
The Spectralis is a spectral-domain OCT (SD-OCT) from Heidelberg Engineering GmbH. The OCT1 model attains scanning speeds of up to 40,000 A-scans per second, whereas the latest OCT2 model attains speeds of up to 85,000 A-scans per second. The Standard Edition 11 of the Spectralis offers a circle scan, labeled the “RNFL” scan, in order to measure cpRNFL thickness. The circle scan diameter subtends 12 degrees, estimated to extend across 3.45 mm and consisting of 768 data points. 12 The scan pattern is manually centered over the ONH, though the software will automatically place the circle scan over the approximate center of the ONH (2.6 degrees nasal and 2.1 degrees) for assistance.
In addition to the circle scan, the Standard Edition 11 offers a volume scan, labeled the “ONH” scan, which consists of a 15 degrees × 15 degrees cube comprising 37,376 data points (73 horizontal B-scans, each consisting of 512 A-scans). The volume scan is manually centered and generates a thickness map of the RNFL.
The principal report of the Spectralis circle scan is the RNFL Single Exam Report, 11 which may be created for one eye or both eyes (Fig. 7‑3c). Similar information may also be viewed immediately during the imaging session with the Thickness Profile Tab. The RNFL Single Exam Report displays sector grids (four- and six-sector, symmetric about the vertical axis) and a thickness graph. In the sector grids, the mean RNFL thickness is indicated for each sector, as well as the global mean of all RNFL thickness values for that eye. The thickness graph plots cpRNFL thickness as a function of the position along the optic disc circumference, with 0 degree located at the intersection of the fovea-disc (FoDi) axis to the circle scan, not the horizontal axis.
Sector grids and the cpRNFL thickness graph adhere to a green–yellow–red significance scheme, whereby green color includes values present in 95% of normal subjects, yellow color indicates values present in the bottom 1 to 5% of normal subjects, and red color indicates values present in the bottom 1% of normal subjects.
An overall classification bar displays a yellow or red color if any sector in either sector grid displays a yellow or red color, respectively. In addition, a confocal scanning laser ophthalmoscopy (cSLO) image with overlaid circle scan pattern and FoDi axis, plus a circle scan tomogram, are available for review.
If a macula scan is also obtained, one may generate a combined RNFL and Asymmetry Analysis Single Exam Report, 11 details of which are found in the following sections.
The Reference Database 11 for the Standard Edition derives from an international cohort and includes 201 healthy subjects of European descent (111 males and 90 females). There is no normative database for the Standard Edition that includes a US population. The mean age of the European cohort is 48.2 ± 14.5 years, with an age range of 18 to 78 years. Refractive errors ranged from +5.00 D to −7.00 D. All subjects manifested the following characteristics: normal intraocular pressure, normal visual field, normal optic disc appearance, and a lack of history of glaucoma. RNFL thickness was found to decrease with age. Thus, all reference data were adjusted with respect to age.
Numerous studies have evaluated the capabilities of the Spectralis Standard Edition. Overall, the area under the receiver operating characteristic curve (AUC) to distinguish all-stage glaucoma from normal reaches a maximum for global (0.952), inferotemporal (0.947), inferior (0.905), superior (0.925), and superotemporal (0.911) cpRNFL thickness. 13 The AUC to distinguish mild glaucoma from normal reaches a maximum for global (0.895), inferotemporal (0.888, 0.858), inferior (0.861, 0.855), and superotemporal (0.873, 0.810) cpRNFL thickness. 13 , 14
Glaucoma Module Premium Edition
The Glaucoma Module Premium Edition 15 (GMPE) offers a novel “ONH-RC” scan (Fig. 7‑4e). This scan consists of three circle scans of estimated diameters 12, 14, and 16 degrees (estimated to be 3.5, 4.1, and 4.7 mm, respectively), and 24 radial scans of length 15 degrees, all automatically centered on the optic nerve using the Anatomic Positioning System (APS) (see below). 16 Each circle scan and each radial scan consist of 768 data points. The GMPE also offers the standard circle scan and volume scan available in the Standard Edition.
One unique component of the GMPE is the APS. 15 Using the APS, the ONH-RC scan itself, as opposed to the postscan analysis grid, is fixed to two structures: the foveal center and the Bruch’s membrane opening center (BMOC). All ONH-RC scans are thus automatically centered on the optic disc based on the BMOC. In addition, a line is drawn to connect these two centers to create an axis along which all scans are aligned, termed fovea-to-Bruch’s membrane opening center axis (FoBMOC). Because the ONH-RC scan consists of circle and radial scans (i.e., not a volume scan), small errors in alignment across longitudinal scans may lead to discrepancies in RNFL measurements. For this reason, proper alignment is paramount and accordingly ensured by the APS. Of importance, note that this differs from the Standard Edition circle scan, which is manually centered—thus inter-scan alignment cannot be guaranteed.
GMPE analytics demonstrate both similarities and differences to those in the Standard Edition. The RNFL Thickness Tab and the corresponding RNFL Thickness Report 15 display sector grids (four- and six-sector) and a thickness graph, as in the Standard Edition. Data displayed may stem from any of the three circle scan diameters (though the inner circle is the default). However, the sector grids adhere to the Garway-Heath division, differing from those used in the Standard Edition (Fig. 7‑3e and Fig. 7‑2f). In this model, the temporal sector subtends 90 degrees (315 to 45 degrees), the superior-temporal sector subtends 40 degrees (45 to 85 degrees), the superior-nasal sector subtends 40 degrees (85 to 125 degrees), the nasal sector subtends 110 degrees (125 to 235 degrees), the inferior-nasal sector subtends 40 degrees (235 to 275 degrees), and the inferior-temporal sector subtends 40 degrees (275 to 315 degrees).
For the sector grid, the mean RNFL thickness is indicated for each sector, as well as the global mean of all RNFL thickness values for that eye. The thickness graph displays RNFL thickness as a function of the position along the optic disc circumference, with 0 degree located at the intersection of the fovea-disc (FoBMOC) axis to the circle scan, not the horizontal axis.
Sector grids and the thickness graph adhere to an identical significance color scheme as the Standard Edition. The remainder of the data and layout are similar to that in the Standard Edition. Of note, a Hood Report may be generated, the cpRNFL specifics of which are discussed later in Section 7.1.4, Analysis.
An RNFL thickness normative database 15 is available for the United States version of the GMPE and thus deviation maps and normative comparisons may be created with cpRNFL data. The US normative database includes 330 eyes of 330 normal subjects (146 male and 184 female) with a racial and ethnic composition representative of the US population and age range of 20 to 90 years. Inclusion criteria included healthy eyes without prior intraocular surgery (except cataract surgery or LASIK); an absence of clinically significant vitreoretinal disease, optic nerve disease, or history of glaucoma; intraocular pressure ≤21 mmHg; best-corrected visual acuity ≥0.5; refraction range +6.00 D to −6.00 D; astigmatism ≤2.00 D; normal visual fields in a glaucoma hemifield test and mean deviation within normal limits; and a clinically normal appearance of the optic disc. As RNFL thickness was found to decrease with increasing age and decreasing BMO area, all reference data were adjusted with respect to age and BMO area.
Numerous studies have evaluated the capabilities of the Spectralis GMPE cpRNFL thickness metrics, chiefly using the area under the receiver operating characteristic curve (AUC), which are listed parenthetically. The capability of the GMPE to distinguish all-stage glaucoma from normal reaches a maximum for global (0.954, 0.86) and inferotemporal (0.929) cpRNFL thickness metrics. 16 , 17 Its capability to discriminate mild glaucoma from normal peaks for global (0.76) cpRNFL thickness and that for preperimetric glaucoma peaks for inferonasal (0.860), global (0.839), inferotemporal (0.767) cpRNFL thickness. 16 , 17 The AUC to distinguish all-stage glaucoma from normal in myopic eyes remains excellent using the global cpRNFL thickness. 18
7.1.3 Avanti RTVue XR (Optovue, Inc., Fremont, CA, USA)
The Avanti RTVue XR is an SD-OCT from Optovue, Inc. Specifications 19 include scanning speeds of 70,000 A-scans per second and the inclusion of two scans: a volume scan (labeled “3D Disc”) and a circle + radial scan (labeled “ONH”). The volume scan images a 6.0 × 6.0 mm area, manually centered over the ONH, and consists of 51,813 data points (101 horizontal B-scans consisting of 513 A-scans). The circle + radial scan is a manually centered scan and consists of 13 circle scans of increasing diameter and 12 radial scans. The radial scans are of length 15 degrees and each comprise 455 data points. The circle scans have diameters ranging from 1.3 to 4.9 mm (increasing by 0.3 mm increments) and comprise 425 to 965 data points, depending on the circle diameter.
The principal RNFL thickness report is the ONH Report 19 (Fig. 7‑5c), which displays thickness values calculated along a 3.4-mm calculation circle (i.e., the cpRNFL). The report displays sector grids of various sizes (two-hemisphere, four-quadrant, and eight-sector, symmetric about the vertical and horizontal axes) with their respective cpRNFL thickness values. The eight-sector display also includes an enclosed thickness map. The average global, superior, and inferior cpRNFL thickness values, plus the difference between superior and inferior cpRNFL thickness, are displayed in a data table. In addition, a thickness graph plots cpRNFL thickness as a function of the position along the optic disc circumference, with 0 degree located along the horizontal axis.
Sector grids and the thickness graphs adhere to a green-yellow-red significance scheme, whereby green color includes values present in 95% of normal subjects, yellow color indicates values present in the bottom 1 to 5% of normal subjects, and red color indicates values present in the bottom 1% of normal subjects.
A cSLO image with overlaid thickness map and circle scan (corresponding to the diameter used for cpRNFL thickness measurements), plus a representative tomogram of the aforementioned circle scan, are available for review. If circle + radial scans are obtained for both eyes, an OU Report may be generated, which additionally displays inter-eye differences of various metrics.
The RTVue Normative Database 19 for the circle + radial (“ONH”) scan of the Avanti RTVue XR consists of 649 eyes from 366 subjects, with a mean age of 49.9 and an age range of 19 to 84 years. Ethnicity percentages included Caucasian (34%), Asian (22%), Hispanic (12%), African Descendant (19%), Indian/Middle Easterner (12%), Pacific Islander (0%), and other (1%). Mean sphere was −0.46 ± 1.9 D (range −7.75 D to +5.50 D).
Inclusion criteria included: age ≥18, refractive error of +8.00 D to −8.00 sphere and +2.00 to −2.00 D cylinder, and best-corrected visual acuity of 20/30 or better in each eye. Exclusion criteria included: a history of leukemia, AIDS, dementia, or multiple sclerosis; concomitant use of hydroxychloroquine or chloroquine; a family history of glaucoma among first-degree relatives; intraocular pressure of ≥22 mmHg or greater in either eye; a visual field demonstrating positive results (>30% false-negative responses, >30% false-positive responses, or >30% fixation losses, pattern standard deviation of p < 5% or worse, and/or a glaucoma hemifield test outside normal limits); active ocular disease including degenerative myopia; previous diagnosis of glaucoma or glaucoma suspect; congenital intraocular surgery or laser treatment (other than refractive surgery or uncomplicated cataract surgery greater than 6 months prior); and/or an anatomically narrow angle.
Correlational analyses revealed that decreased RNFL thickness was associated with increasing age and decreasing optic disc size. For this reason, all clinical data comparisons to the normative database are adjusted for age and optic disc size.
As measured by the area under the receiver operating characteristic curve (AUC, listed parenthetically in the following section), the capabilities of the Avanti SD-OCT to distinguish all-stage glaucoma from normal reach a maximum for average (0.968, 0.919, 0.87, 0.879) and inferior (0.947, 0.884, 0.86) cpRNFL thickness. 2 , 3 , 20 , 21 The capability to distinguish mild or preperimetric glaucoma from normal remains excellent though slightly decreased compared to that for all-stage glaucoma. 3 , 22 Of note, in highly myopic eyes, the diagnostic capability of cpRNFL has been shown to decline compared to emmetropic eyes. 4
7.1.4 3D OCT (Topcon Corporation, Tokyo, Japan)
The 3D OCT series from Topcon Corporation includes the 3D OCT-2000 and the 3D OCT-1 Maestro2, the latter being the latest model in this series. The OCT-2000 23 attains scanning speeds of 50,000 A-scans per second. The OCT-2000 offers two scans: a volume scan, labeled “3D 6.0×6.0,” and a circle scan, labeled “Circle 3.40.” The volume scan consists of a manually centered 6.0 × 6.0 mm scan consisting of 65,536 data points (128 horizontal B-scans, each consisting of 512 A-scans). The volume scan derives a peripapillary calculation circle of 3.4 mm diameter. The standalone circle scan consists of a 3.4-mm-diameter circle with 1,024 data points and is automatically centered using the optic disc auto-search function, which identifies the optic disc center through registration to both infrared and color fundus photographs.
The 3D OCT-1 Maestro2 24 is the latest SD-OCT model from Topcon Corporation. The device attains scanning speeds of 50,000 A-scans per second. The OCT-1 Maestro2 offers a single scan: a manually centered 6.0 × 6.0 mm volume scan consisting of 65,536 data points (128 horizontal B-scans, each consisting of 512 A-scans). The software derives a peripapillary circle of 3.4 mm diameter composed of 1,024 data points and automatically centered on the optic disc using the retinal pigment epithelium (RPE) edge as the optic disc margin.