Making the diagnosis of glaucoma requires the presence of optic nerve head (ONH) damage and/or characteristic visual field changes. Although automated and static perimetry provides an excellent aid in documenting functional visual field loss from glaucoma, as much as 40% of the retinal nerve fiber layer (RNFL) may be lost before changes are detected on standard white-on-white visual field testing.¹ The need for more sensitive, reproducible, and accurate detection of glaucomatous damage has led to the development of various imaging devices for ONH and RNFL. To best utilize these new technologies, it is essential that physicians familiarize themselves with the information the tests provide and the strengths and weaknesses of each modality (Table 6-1).

Assessment of changes in the ONH and RNFL over time can be evaluated using color stereophotographs (Fig. 6-1). This method is the most widely used imaging technique and is considered the gold standard for documentation of glaucomatous optic neuropathy (GON).¹

Stereophotographs can be produced by taking two photographs in sequence, either by manually repositioning the camera or by using a sliding carriage adapter (Allen separator). Alternatively, they can be produced by taking two photographs simultaneously with two cameras that utilize the indirect ophthalmoscopic principle (Donaldson stereoscopic fundus camera) or a twin-prism separator. Simultaneous ONH photographs have demonstrated the best reproducibility over time.² Recently, alternation flicker of monoscopic optic disc images over time (digital stereochronoscopy) has also proved to be a sensitive method for detecting GON changes. Newer image processing and registration permits accurate alignment of optic nerve (ON) photographs and facilitates image comparison with detection of changes in vessel position, color, and other cues for contour change.³ Digital stereochronoscopy may assist in optimizing the ability to detect structural changes and overcome some of the assessment variability between observers.⁴

The strengths of ONH stereophotographs include the ability to document parameters that cannot be quantified, such as disc hemorrhages, peripapillary atrophy, and pallor. They also allow the clinician to assess the impact of other nonglaucomatous processes that may influence functional testing. However, stereophotographs provide excellent documentation of the ONH, but their interpretation remains subjective and can therefore have increased variability and limited usefulness over time.⁵ In addition, media opacities, such as cataracts, or a poorly focused photograph can inhibit optimal analysis.

Heidelberg retina tomography (HRT; Heidelberg Engineering GmbH, Heidelberg, Germany) is a type of confocal scanning laser ophthalmoscopy used to provide a structural quantitative measurement of the ONH (Fig. 6-2A). A 670-nm diode laser is aimed through a pinhole onto the retina. The light reflected passes through a second pinhole into a detector, which transfers the maximum intensity of light at a given point to create an image. A series of 16 to 64 two-dimensional (2D) sequential scans, each measuring 384 × 384 pixels, is used to create a three-dimensional (3D) color-coded topographic image (Fig. 6-2B).

The reflectivity image (Fig. 6-2C) produced is used by the technician to draw a contour line around the inner border of the scleral ring, which is further separated into six sectors. This line remains standard for all future examinations and is used to calculate a series of parameters. The HRT 3.0 reduces technician dependence by producing automated results of the contour line as described by Swindale et al.¹ The graphic analysis (Fig. 6-2D) corresponds to all six sectors in the reflectivity image.

A reference plane defined as 50 µm below the surface of the retina along 6 degrees of the contour line in the temporal inferior region is designated as the cutoff between the neuroretinal rim (all structures above the reference plane) and the cup (all structures below the reference plane).

After the data are reconstructed, 12 stereometric parameters (Fig. 6-2E) are automatically obtained. According to univariate and multivariate analysis from the ocular hypertension treatment study, the most predictive parameters for glaucoma detection with the HRT were the ones independent of the reference plane (mean height contour, rim area, and mean cup depth).² Topography standard deviation is used as a measurement of image quality and reliability, with anything greater than 30 to 40 µm considered unreliable. Moorfield’s regression analysis (Fig. 6-2D) uses linear analysis between optic disc area and neuroretinal rim area in both global and localized segments to classify the nerve as “within normal limits,” “borderline,” or “outside normal limits” based on comparison to an age- and ethnicity-specific normative database.

Progression of GON on the HRT can be evaluated with event- or trend-type analysis software. Topographical change analysis (Fig. 6-3) is an event-type analysis that focuses on the difference between surface height measurements of follow-up images and those of a baseline image. Changes that are consistent are highlighted in a color-coded overlay on top of the reflectivity image, with red representing areas of statistically significant worsening. Trend-type analysis assesses the rate of change over time in one of the stereometric indices, such as rim area, to document progression.

Several studies have confirmed the ability of HRT to distinguish between healthy and glaucomatous eyes with a range of specificity from 75% to 95% and sensitivity from 51% to 97%.³,⁴,⁵ Reproducibility is enhanced when at least three scans are combined.⁶

Among advantages of the HRT is the company’s commitment to assure older software remains compatible with newer machines, thereby allowing for longer data series on individuals. Additional benefits include good image quality through undilated pupils and the capability to assess glaucomatous progression by comparing sequential scans, using either the glaucoma change probability analysis or topographic change analysis software.⁷,⁸ A disadvantage of the older HRT 2.0 model is the requirement for an operator-dependent outline of the disc margin. An additional limitation is the standardized reference plane used to calculate many of the parameters because this may not accurately represent the true RNFL thickness in all individuals. These concerns have been addressed with the newer HRT 3.0 model. In addition, the HRT 3.0 software contains a larger ethnicity-specific normative database.⁹