Glaucoma is a characteristic optic neuropathy characterized by retinal ganglion cell death, which then leads to retinal nerve fiber layer (RNFL) thinning and optic nerve head (ONH) cupping. The most important risk factor is elevated intraocular pressure. Since these structural changes in the RNFL and ONH may result in irreversible visual field (VF) loss,^8,9 the early diagnosis of glaucoma is vital for the early initiation of treatment that may stop or slow down further permanent vision loss. One reason why early diagnosis often does not occur is that patients are usually asymptomatic early in their disease. Another problem preventing early diagnosis is that the standard clinical methods for diagnosing glaucoma (ie, VF testing) can only diagnose glaucomatous vision loss after up to 40% of the nerve tissue is lost irreversibly.^10–13 VF testing is also not an ideal method to diagnose glaucoma because these tests rely on not only the subjective test response of the patient but also the subjective interpretation of the physician.

Because of the need to diagnose glaucoma not only earlier but also in a more objective manner, various imaging technologies have been developed over the past decade in an attempt to quantitatively analyze progressive glaucomatous structural changes, such as RNFL thinning and/or ONH cupping, both of which are known to precede clinically detectable VF defects. So far, there are 3 main commercially available imaging technologies that are used to image glaucomatous structural changes: confocal scanning laser ophthalmoscopy (CSLO) (Heidelberg Retina Tomograph—HRTII and III, Heidelberg Engineering, Germany), scanning laser polarimetry (SLP) (GDxVCC, Carl Zeiss Meditec Inc, Dublin, CA), and optical coherence tomography (OCT). We feel that spectral domain optical coherence tomography (SDOCT), also called Fourier domain OCT, has the greatest potential for imaging glaucomatous structural changes.

SLP imaging is possible if a structure is birefringent, and birefringent structures in the eye include the cornea and the RNFL. The RNFL exhibits birefringence owing to the orderly arrangement of axons and microtubules within the RNFL. When the GDx sends a 780-nm polarized laser beam through the RNFL, phase retardation of the polarized light occurs. A simplified formula [phase retardation = birefringence × RNFL thickness × (2π/wavelength)] shows that the amount of phase retardation is proportional to RNFL thickness. The GDx then calculates RNFL thickness by taking the phase retardation, the actual value that is measured by the GDx machine, and dividing it by a constant birefringence value. Both the GDxFCC and GDxVCC (formerly of Laser Diagnostic Technologies Inc, San Diego, CA) and the GDxVCC (now of Carl Zeiss Meditec Inc, Dublin, CA) assume a constant RNFL birefringence value for all patients and for all areas of the retina. Studies have since shown that RNFL birefringence values in normal subjects not only are different for different people but also may vary in different areas of the retina, and RNFL birefringence values change with glaucomatous damage.^16–18 As direct in vivo measurements of RNFL birefringence values vary 3-fold in normal patients, RNFL thickness values may be incorrectly calculated by the GDxFCC and VCC machines by over 30%.¹⁶

Unlike CSLO and SLP, OCT can directly image both the ONH and the RNFL. Unlike ultrasound imaging where sound travels more slowly, light travels so fast that a fundamentally different way to process information from the eye is needed in OCT. For the leading commercially available TDOCT machine, Stratus OCT Model 3000 (Carl Zeiss Meditec Inc, Dublin, CA), the laser in the “source arm” is a 820-nm low-coherence near-infrared superluminescent laser diode (SLD) light source (Fig. 3). The light from the SLD light source passes through a fiber coupler that splits light to the “reference arm” and the “sample arm” (ie, the eye). The interference from light coming back from the reference arm and the eye is then received by a photodetector in the “detector arm.” By translating the reference arm mirror and spinning the scanning galvanometer mirror, a 2-dimensional image of the scanned optic nerve or retina is created. The axial resolution of the Stratus OCT is approximately 10 μm.

For ONH imaging, the Stratus OCT scans along 6 radial 4-mm line scans centered on the ONH (Fig. 4). Interpolation is used to fill in the missing data about the ONH topography between the 6 radial line scans. As the Stratus OCT takes over 1 full second to scan a 2-dimensional image of the ONH, the scan area is limited and microsaccades and motion artifacts are introduced. As the latter necessitates realignment of the A-lines, true ONH topography with TDOCT is not possible.

RNFL measurements are acquired along a 3.46-mm diameter circular scan around the ONH and are presented graphically as a “double hump” in the normal eye (Fig. 5). However, the Stratus OCT does not measure the RNFL thickness either inside or outside this circular scan.

Most importantly, OCT provides clinically relevant images that correlate well with histology and has thus been proven to be an important technology in the evaluation of glaucoma and retinal diseases.^20–22 Despite the advantages of OCT technology, the main limitations of the commercially available TDOCT machine are its low resolution (ie, 10 μm) and slow acquisition speed (ie, average 1.28 s for a 2-dimensional image composed of 512 A-lines).

To achieve ultrahigh resolution (UHR) TDOCT imaging, which enables better axial resolutions of <3 μm, a titanium:sapphire laser source has been used in prototype machines, which are not commercially available. However, there are 2 disadvantages of the UHR TDOCT imaging system: (1) the titanium:sapphire laser costs almost 10 times the cost of the Stratus OCT3’s SLD light source and (2) the higher resolutions are achieved at the expense of slower acquisition speeds, which is a consequence of the inverse relation between resolution and acquisition speed.^23–25

Despite the advancements in CSLO, SLP, and TDOCT technology, none of these methods have proven to be better than a complete eye examination with stereo disc photography and VF testing.^14,15 Clinical studies have suggested that the sensitivities and specificities range from 74% to 94% and 65% to 91%, respectively, for detecting glaucoma using these 3 technologies.^14,26–31 Certain abnormal parameters for all of these 3 technologies in glaucoma suspects have been shown to be associated with the future development of primary open angle glaucoma.^32–34

The first prototype video-rate SDOCT system was built by Johannes de Boer, PhD, at the Massachusetts General Hospital.^36–42 Although the light source is the same as in TDOCT (ie, either an SLD light source or a titanium:sapphire laser source), the main difference between TDOCT and SDOCT is the way the information is processed as light comes back from the reference arm and the eye (Fig. 6). In contrast to TDOCT, which uses a point detector or photodetector in the detector arm (Fig. 3), SDOCT uses a spectrometer (Fig. 6) that is composed of the transmission grating and the air-spaced focusing lens. In SDOCT, depth information is acquired by analyzing the interference patterns in a spectrum of mixed reflected lights.^25,35,43,44 This information from the spectrometer then undergoes Fourier transformation to create an image; therefore, this technology has also been referred to as Fourier domain OCT.^25,36

Because of the greater efficiency of the spectrometer, ultrahigh speed data acquisition of weaker signals is possible. SDOCT’s higher acquisition speeds allow for the transition from 2-dimensional to 3-dimensional and video ophthalmic imaging (Fig. 7). With SDOCT, A-lines can be acquired at a rate of 29,000/s (at 6-μm axial resolution), which is 73 times faster than the Stratus OCT3. At a 14,500 Alines/s rate, a 3.5-μm axial resolution can be achieved, which is 3 times better than Stratus OCT3.^36,41 Since SDOCT can acquire 29 frames/s (1000 A-lines per frame), a 2-dimensional image can be obtained in 1/29 of a second. Because faster acquisition speeds allow for scanning of larger areas of the retina, SDOCT can create RNFL thickness maps of large regions of the posterior pole (Fig. 8).

Unlike TDOCT, which achieves UHR images by increasing acquisition times, SDOCT can achieve UHR imaging near 2-μm axial resolution without a significant increase in acquisition times and with a titanium:sapphire laser source (Figs. 9, ​,1010).^42,45 The FDA-approved commercially available SDOCT machines to date only give us axial resolutions of about 5 to 7 μm (compared with Stratus TDOCT at 10 μm), because these machines use the cheaper, easier-to-maintain SLD light source.

The image quality, expressed as the SNR (signal-to-noise ratio), of the SDOCT system is better than that of TDOCT.^37,38 For example, in the experimental prototype SDOCT system that was at the Massachusetts Eye and Ear Infirmary, the SLD light source was centered at 840 nm with a bandwidth range of 50 nm. The axial resolution in a TDOCT system would increase with optical bandwidth, but its SNR is inversely proportional to an increase in optical bandwidth.^41,42 As the light source bandwidth does not affect the SNR performance of an SDOCT configuration, any source can be used in SDOCT without compromising image quality. Unlike titanium:sapphire laser sources, ultrabroad bandwidth superluminescent diode light sources have been developed that are compact, relatively inexpensive, and require low maintenance. The combination of new light sources and SDOCT technology has greatly improved ophthalmic imaging. In addition, the SNR can be reduced further with pulsed illumination instead of continuous wave illumination.^46–48 In analogy to stroboscopic illumination, sample motion can be frozen by using a pulsed light source.⁴⁷ This pulsed illumination subsequently reduces the detrimental effects of sample motion during scanning, thereby providing a 4.4 relative SNR advantage compared with that of continuous wave illumination.⁴⁸ As a result, less artifacts and clearer images can be acquired. SDOCT has, therefore, evolved into a preferred technology for UHR ophthalmic imaging.

By the end of the year 2006, many imaging companies began to market SDOCT machines in the United States, and some of these machines and companies include the following: RTVue (Optovue, Fremont, CA), Spectralis (Heidelberg Engineering Inc, Heidelberg, Germany), SOCT Copernicus (Optopol Technology, Zawiercie, Poland), Cirrus HD-OCT (Carl Zeiss Meditec Inc, Dublin, CA), 3D OCT-1000 (Topcon, Paramus, NJ), etc. At the time of this writing, not all of these machines have glaucoma software or an age-matched normal database.

Because the new SDOCT machines give 3-dimensional data instead of TDOCT’s 2-dimensional data, new algorithms for SDOCT machines need to be developed to better evaluate the ONH in glaucoma patients. For ONH imaging and analysis, SDOCT can produce a 2-dimensional integrated reflectance image, which is similar to a CSLO image (Fig. 8, top). Although CSLO and TDOCT have been largely limited to ONH topography and 2-dimensional imaging, SDOCT may now allow us to better measure neuroretinal rim tissue volume. A parameter called the “minimum distance band” (MDB) may better reflect the actual amount of nerve tissue coursing through the ONH (Fig. 11).⁴⁹ In this parameter, the “minimum distance” refers to the band of tissue or the shortest distance between the retinal pigment epithelium edge and the ONH surface. All nerve axons coming from the photoreceptors that course through the ONH on the way to the brain must necessarily pass through this narrow MDB, which encircles the cup. This band area could provide an objective quantification of the nerve tissue in the neuroretinal rim, and any focal regions of thinning in this band may indicate glaucomatous neuroretinal rim thinning. As determination of the MDB or neuroretinal rim area only requires determination of the highly reflective ONH surface and the also highly reflective retinal pigment epithelium edge, this value should be easily determined in all glaucoma patients. This ability to consistently determine an objective 3-dimensional parameter of nerve tissue becomes more significant when we realize that not all glaucoma patients can generate a reliable RNFL thickness map.

Because ultrahigh acquisition speeds now allow for 3-dimensional scanning of large areas of the retina, SDOCT can create RNFL thickness maps of large regions of the posterior pole. This is in contrast to TDOCT Stratus OCT that only determines the RNFL thickness values along a circular scan around the ONH, but does not give any information of RNFL thickness either inside or outside this circular scan (Fig. 12). The importance of RNFL thickness determination in the early diagnosis of glaucoma can not be overstated because RNFL thinning may in theory be the earliest structural change clinically detectable as RNFL thinning has been shown to occur as early as 6 years before VF loss.¹¹ RNFL thickness maps can also be potentially used for a thorough evaluation of the RNFL in the longitudinal monitoring of glaucoma patients, which is extremely important in the clinical management of a life-long disease.⁵⁰ The problem with relying on RNFL thickness measurements in evaluating glaucoma patients is that this parameter is more difficult to reliably measure compared with ONH topography. When determining RNFL thickness for SDOCT RNFL thickness maps, the anterior and posterior boundaries of the RNFL can be determined owing to the reflectivity changes between the RNFL-vitreous and RNFL-inner plexiform layer.⁵⁰ When the RNFL undergoes glaucomatous changes and thinning, the reflectivity of this tissue decreases, which makes the posterior RNFL border often difficult to determine. However, ONH topography is always very easy to determine in OCT because of the consistently sharp change in reflectivity between the vitreous and the ONH surface, even if the ONH undergoes glaucomatous cupping. Therefore, ONH topography and new SDOCT ONH parameters (ie, MDB) may be more consistently reliable parameters in the evaluation of glaucoma patients.