Ординатура / Офтальмология / Английские материалы / Shields Textbook of Glaucoma, 6th edition_Allingham, Damji, Freedman_2010

1 - Cellular and Molecular Biology of Aqueous Humor Dynamics Page 175 of 225

relationship between visual perception and a change in light intensity. A decibel is 0.1 log-unit, so that a 10 dB represents a 10-fold decrease of the maximum stimulus of any specific perimeter, and a 20 dB represents a 100-fold stimulus attenuation. The maximum intensity of a perimeter has a value of 0 dB, meaning that the stimulus is not attenuated. Log-units and decibels are relative units, and resulting stimulus intensity is not the same for all instruments, but decibels represent the same percentage change of the intensity in all perimeters.

Stimulus Size

The standard target for kinetic and static perimetry is a white disc, the stimulus value of which can be adjusted by varying the target size or luminosity relative to that of the background. In healthy persons, the mean retinal sensitivity has been shown to increase with the increasing size of the test object (104). If the diameter of the smaller stimulus is increased, it may be as visible as the less intense larger stimulus, the phenomenon known as spatial summation. Usually, doubling the stimulus diameter has the same effect on the visibility of the stimulus as increasing its intensity by 5 dB (1).

Exposure Time

The exposure time will also affect the stimulus visibility. The stimulus presented over a longer period of time may become more visible, the phenomenon called temporal summation. However, after the temporal summation is complete, which happens typically after 0.1 second, the image is not seen any better. The Humphrey field analyzer uses a 0.2-second stimulus duration, which also helps prevent movement of the patient's gaze toward the stimulus. However, suprathreshold static targets should be presented for a longer time, usually 0.5 to 1 second, and test objects should be just above threshold for the area being tested.

Kinetic versus Static Perimetry

The threshold is theoretically the target that is just bright enough to be seen 50% of the time at that location (the differential light threshold). The stimulus that is below the threshold value cannot be seen. Kinetic perimetry defines threshold by moving the test object from a nonseeing (subthreshold) to a seeing (suprathreshold) area, and by recording the point at which it is first seen in relation to fixation (Fig. 5.10A). The procedure documents the boundaries of the visual field for the absolute limits and areas of relative differences in visual acuity within the field (Fig. 5.11). As previously noted, the boundaries, or contour lines, are called isopters. The size and shape of a particular isopter depend partly on the stimulus value of the corresponding test object.

Static perimetry involves the presentation of stationary test objects, by using suprathreshold or threshold presentations. Suprathreshold static presentation is an “on-off” technique in which a test object just above the anticipated threshold for the corresponding portion of the visual field is momentarily presented, and the points at which the patient fails to recognize the target are noted as visual field defects. It is a way of “spot checking” for areas o f relative or absolute blindness, usually in the central visual field. The suprathreshold strategy is used mostly as a screening test.

Threshold static perimetry measures the relative intensity thresholds for the visual acuity of individual retinal points in the field of vision. The technique involves gradually increasing the target light from subthreshold intensity in small increments, and recording the level at which the patient first indicates recognition of the target (Fig. 5.10B), or decreasing

P.101

it from a suprathreshold level and recording the lowest stimulus value seen. The points are tested at predefined locations throughout the visual field, and the results are recorded as grayscale symbols and numerical sensitivity values in decibels (Fig. 5.12).

Figure 5.10 Standard techniques for measuring the visual field. In kinetic technique (A), test object moves from nonseeing to seeing area. Static technique (B) measures sensitivity of retina at a given point.

Figure 5.11 Example of manual kinetic perimetry showing two complete isopters (l2 and l4) and a third partial isopter (V4) in nasal periphery with blind spot measured by l2 target.

The kinetic stimuli are usually seen better than the static ones are, but when the stimulus is moved slowly, the results of kinetic and static perimetry are similar.

To minimize the patient's anticipation of when or where the next test object will appear, the presentation should be random, rather than following a predictable pattern, and the time between stimuli should be varied slightly. To avoid patient anxiety when testing in a nonseeing area, the examination should return periodically to a previously seen area.

Figure 5.12 Examples of threshold static (standard automated) perimetry. Retinal sensitivity is measured at points throughout a portion of the visual field (central 24 to 30 degrees in this example). Results can be displayed in numerical values and symbols.

For kinetic targets, a stimulus velocity of 4 degrees/sec appears to be optimal for all targets in the central and peripheral visual field, but a slower velocity of 2 degrees/sec may provide more reproducible results in some patients (105, 106). The test object should always be moved from a nonseeing to a seeing area—that is, from the periphery toward fixation wh en outlining an isopter and from the center of the blind spot or a scotoma.

P.102

Threshold static perimetry has been shown to be more sensitive than kinetic perimetry is in detecting glaucomatous field loss (107, 108). In one study of patients with COAG, a defect was found in one third of the cases with static perimetry that was missed by kinetic perimetry (109). In a long-term study of patients with ocular hypertension, 75% of those who developed glaucomatous damage had an abnormality detected by automated static perimetry (by using a hemifield test, explained later) 1 year before field loss was detected by manual perimetry, by using a combination of kinetic and static presentations (110).

When automated static perimetry was compared with Goldmann kinetic perimetry as a test for driving, a significant number of patients with severe field defects, detected by static perimetry, still met the standard for driving by the kinetic perimetry (111).

Because standard static threshold perimetry tests sensitivity near threshold, patients do not see approximately half the presented stimuli, and they may report that stimuli are too dim to see. Patients should be told that the limits of their seeing abilities are being tested and that barely seeing the stimuli is natural.

Background Illumination

Background illumination for manual perimetric techniques traditionally stimulates both rods and cones. The adapting field luminance currently used in static and kinetic perimetry is marginally photopic (e.g., 31.6 apostilbs), although the optimum luminance has yet to be established. One study suggested that the lower levels of background illumination may allow minor reductions in light transmission by the ocular media to produce significant changes in the recorded threshold sensitivity (112). In a comparison of scotopic and photopic fields, localized scotomas in patients with glaucoma were of equal depth, but diffuse scotopic defects significantly exceeded the photopic, supporting the concept that not all ganglion cell types are equally susceptible to glaucomatous damage (113). Scotopic defects were also found more often in patients with ocular hypertension or glaucoma than in healthy persons, and the defects were mainly in the superior hemifield (114).

With bowl perimeters, photometric adjustment should be made with the patient in place, because facial coloring affects luminosity. The most important principle regarding illumination is to keep the target and the background constant and reproducible from one examination to the next.

Physiologic Influences on Visual Fields

The following factors should be compensated for, if possible, or otherwise should be considered when interpreting the fields.

Pupil Size

Although decreased pupil size should have little effect on a patient's perception of a stimulus, because background and stimulus are affected equally, significant miosis may depress central and peripheral threshold sensitivities and exaggerate field defects (115), even after correction of induced myopia (116). One study used neutral density filters to reduce the retinal illumination by the equivalent of halving the pupillary diameter, which reduced the mean threshold with two automated perimeters by 1.1 to 1.7 dB (117). In another study, use of pilocarpine worsened the visual field global indices, such as mean deviation and pattern standard deviation (explained later) (118). For this reason, the pupil size should be recorded with each field, and the influence of miosis should be considered when a field change is detected. Mydriasis has less influence on the visual field than miosis does, although pupillary dilatation with use of tropicamide, 1%, or no ocular medication in healthy persons reduced threshold sensitivity with automated perimetry in one study (119).

Age

Increasing age is also associated with reduced retinal threshold sensitivity (120). This effect starts as early as 20 years of age, progresses linearly throughout life, and involves the peripheral and superior areas more than the pericentric and inferior portions of the field (121, 122). This age-related visual field sensitivity appears to be primarily due to neural loss rather than preretinal factors (123). Standard automated perimetry (SAP) protocols compensate for the effect of age by using age-bracketed databases. Clarity of Ocular Media

Cataracts produce glare and change the intensity of the stimulus. Therefore, a cataract can cause or exaggerate central or peripheral field defects, which could be mistaken for the development or progression of glaucomatous field loss. Even minimal light scattering, as may be caused by an early cataract that has a relatively insignificant effect on visual acuity, may influence threshold measurements (124). As previously noted, this effect may be greater with lower levels of background illumination (112). Eyes with COAG and cataracts may have improvement of foveal sensitivity, visual field scores, and sometimes even a reversal of a partial or complete scotoma after cataract extraction (125, 126 and 127). However, cataract surgery can also reveal mild and moderate field defects masked by cataracts (128, 129). Nuclear cataracts depress central perimetric sensitivity more than peripheral sensitivity with both large and small targets, whereas nonnuclear cataracts influence central sensitivity more for small targets and peripheral sensitivity more for large targets (130). Attempts have been made to correlate visual field damage with lens opacity and visual acuity to aid clinicians in determining the significance of field change in patients with glaucoma and cataracts (131, 132).

Reduced clarity of the ocular media from other causes, such as a corneal disturbance, a cloudy posterior lens capsule after cataract surgery, or vitreous opacities, may also affect the visual fields. Applanation tonometry before automated static threshold perimetry was found to have no detrimental effect on the visual field results (133).

Refractive Error and Retinal Blur

When the projected stimulus is not focused on the retina, the edge of the stimulus is blurred, contrast is decreased, and the stimulus may not be detected by the patient. The larger

P.103

the stimulus, the less it is to be affected by the blur. Refractive errors primarily influence the central field (134). When a standard size III stimulus is used, refractive errors of 1 diopter (D) or less may not need to be corrected, because it usually will cause only slightly more than 1 dB of general reduction of sensitivity (135). Mild myopia requires no correction, unless the refractive error exceeds 3 D. Posterior

staphylomas can create areas of relative myopia, called refraction scotomas, which may be confused with glaucomatous field defects, but can usually be eliminated with an appropriate refractive correction. Hyperopia has a greater influence on perimetric results, especially for the central field, and even small hyperopic refractive errors can significantly alter threshold sensitivity (134, 135 and 136). Age tables are available to aid in determining the appropriate correction for presbyopia. A contact lens provides the best correction for the aphakic and highly myopic eyes (137), although spectacle correction can be used for the central 24 to 30 degrees with no correction for the peripheral field. Astigmatism should be corrected unless the cylinder is less than 1 D, in which case it can be included as the spherical equivalent.

Psychological Influences on Visual Fields

Patients' understanding of the test and their alertness, concentration, fixation, and cooperation all affect the results of visual field testing (138). A learning effect with automated perimetry may influence the results of a patient's first or second field test, suggesting that an initial field that does not agree with the clinical findings should be repeated (139, 140 and 141). One study found that patients with refractive errors, especially those with myopia, had a larger learning effect than patients with emmetropia did (142). Another study found that moderate alcohol intake did not influence differential light sensitivity as tested by automated perimetry (143). With manual perimetry, the skill of the perimetrist influences the visual field test results (144).

Patient Fatigue

Full-threshold protocols take a long time to complete, and patients usually find visual field testing exhausting. Fatigue causes artificially decreased sensitivity in the areas of existent glaucomatous defect (145). Fatigue may also cause decreased performance in patients with glaucoma within central 10 degrees, and increased deterioration of the mean defect and localized loss in the periphery (146, 147). TECHNIQUES AND INSTRUMENTS FOR MEASURING THE FIELD OF VISION

Just as a cartographer maps the boundaries and topography of an island, so the perimetrist can measure both the peripheral limits of a visual field and the relative visual acuity of areas within those limits. This may be accomplished by using static or kinetic techniques with instruments that are computer assisted (automatic) or manually operated.

Automated Static Perimetry

Automated perimetry is accepted as the standard way of measuring the visual field. The standard protocol of static white-on-white stimuli is commonly known as SAP. A major limitation of tangent screens and arc perimeters (discussed later) was lack of standardization of the test objects and the background, and patient fixation. These needs were addressed in the era of standardization, which began in the middle of the 20th century with the contributions of Goldmann. The main problem that remained, however, was the subjectivity of the patient and the perimetrist. Although subj ectivity of the patient has not been eliminated, the influence of the perimetrist was eliminated to variable degrees with the advent of automated perimetry in the 1970s. A wide variety of automated perimeters have been designed since then. Many of these are no longer commercially available, but current models represent modifications of the originals.

By reducing the influence of the perimetrist, automated perimetry improves the uniformity and reproducibility of visual fields. With these instruments, the perimetrist only ensures that the patient understands the testing procedure, is comfortably positioned at the perimeter, and adheres to the requirements of the test. In addition, the use of computers has provided new capabilities that are impossible with manual perimetry, including random presentation of targets, estimations of patient reliability, reduced variability, and statistical evaluation of data at many levels. With the recent introduction of efficient threshold strategies, automated perimetry is not only more accurate and informative but is also faster than manual perimetry.

Basic Components of Automated Perimeters

Automated perimeters have two main components: the perimetric unit and the control unit. The perimetric unit in most systems uses a bowl-type screen, similar to that of the Goldmann manual perimeter (discussed later).

1 - Cellular and Molecular Biology of Aqueous Humor Dynamics Page 181 of 225

The control unit provides interaction between the operator and the computer through a dialogue screen and a keyboard or light pen. The computer in the control unit provides and monitors instrumentation function according to the perimetrist's request, evaluates the patient's response, and processes data. The control unit also contains a printer, which provides a hard copy of the data in symbols and numeric values. Computers also store recorded information and can perform statistical analyses of the data in relation to the programmed normal database, or against previous fields for the same patient.

Static targets are used in most automated perimeters. Automated kinetic targets have also been evaluated and are provided on some automatic perimeters, although rarely used today, probably because of the high frequency of fixation errors and longer testing time (11, 12 and 13, 106, 148, 149). The targets may be projected onto the bowl, which is the current standard, or illuminated from light-emitting diodes (LEDs) or fiber optics in the perimetric bowl in earlier models. The former has the advantage of unlimited presentation locations on the screen, whereas the latter two have fixed positions in the bowl. In addition, the LEDs

P.104

were recessed in dark cavities, which may allow perception by the most sensitive retinal areas of a stimulus that is of lower intensity than the background light (150, 151). This “dark hole phenomenon” is associated with increased variability in retesting the threshold (150, 151). Projected targets also have the advantage of allowing for change in size to alter the stimulus values. In practice, the size is usually kept constant, although larger targets may permit the measurement of visual function in areas that had been considered absolute scotomas with standard-sized stimuli (152). A larger target (size V stimulus) was found to be useful in patients with end-stage glaucoma (153). With all target systems, the patient usually presses a button to indicate when a target is seen, which is recorded by the computer.

The standard stimulus in most automated perimeters is a white light on a white background, which tests the patient's differential light sense.

Commercial Units

The first of the full-threshold perimeters to receive extensive study was called the Octopus. With each Octopus model, stimuli are projected onto a bowl, and fixation is monitored by the corneal light reflex method and a television view of the patient's eye. The models differ primarily according to computer capabilities. These automated perimeters were shown in early studies to compare favorably with manual perimetry and to frequently detect field loss missed with the Goldmann perimeter (154, 155 and 156). The Humphrey field analyzer and Humphrey field analyzer II also use projected stimuli on a bowl (Fig. 5.13). They monitor fixation by the Heijl-Krakau periodic blind spot check method and also by corneal light reflex in newer models. It is currently the most commonly used automated perimeter. It has also compared favorably with manual perimetry on the Goldmann perimeter, often detecting defects that the latter missed (157). In one study, however, patients preferred the Goldmann perimeter, whereas the technician favored the Humphrey (158). The Octopus and Humphrey units have been compared in several studies. In one study, both shortand long-term fluctuations (explained later) were greater with the Octopus (159). In another study, both automated perimeters identified slightly more defects by meridional threshold testing than the Tübingen manu al perimeter did (160) (discussed later).

Figure 5.13 Humphrey Field Analyzer (HVAII). (Courtesy of Carl Zeiss Meditec, Inc.) Test Patterns

A broad menu of test patterns is available with most instruments. The most commonly used are limited to the central 24 to 30 degrees, with a 6-degree separation between test locations. The 6-degree grid may miss the physiologic blind spot and small glaucomatous defects in a high percentage of cases, and it has been suggested that tighter grids should be used, especially in the central 10 to 28 degrees (161, 162 and 163). Special programs are available to study smaller portions of the field with tighter grids. Programs are also available to study the peripheral field beyond 30 degrees in the nasal quadrant or for 360 degrees. The peripheral studies can be performed alone or in conjunction with a central field program and usually have wider target separation. Static testing of the peripheral nasal field has been shown to provide valuable additional information in detecting glaucomatous defects (164). Automated kinetic measurement of the peripheral field, especially nasally, was also found to provide useful information in many patients, in addition to the information obtained from central testing (11, 12 and 13). One study of various factors that affect the reaction time during automated kinetic perimetry led to the suggestion that the test should be designed to adjust to individual patient responses, because other factors, such as eccentricity or luminance level, were found to have much smaller effect on reaction time within the central 30 degrees (106).

Testing Strategies

All fully automated perimeters take advantage of computer capabilities by using random presentation of the static targets to avoid patient anticipation of the next presentation sites. In addition, an adaptive technique is used, in which stimuli are presented according to the presumed normal retinal threshold

1 - Cellular and Molecular Biology of Aqueous Humor Dynamics Page 183 of 225

contour (i.e., the relative differential light thresholds throughout the visual field), on the basis of agecorrected normal data or the patient's response to preliminary spot tests (Fig. 5.14). This approach, in comparison with the presentation of a constant stimulus value throughout a portion of the field, as with many manual techniques, improves the balance between sensitivity (the ability to detect defects) and specificity (the ability to detect normal areas). Fully automated perimeters provide suprathreshold and full-threshold measurements.

Suprathreshold Static Perimetry

Suprathreshold static perimeters present a stimulus brighter than the anticipated normal value for the corresponding retinal location. Some instruments simply indicate whether the target was seen, whereas others present a second, high-intensity target in nonseeing areas to distinguish between relative and absolute defects. In either case, however, these instruments are limited

P.105

to screening functions, in that they do not provide sufficient information about the depth or contour of a field defect to be used as a baseline study or for following up the patient during therapy. With the continued advances in automated perimetry, these suprathreshold strategies have been largely replaced by full-threshold strategies, although suprathreshold models may have value as screening devices. Improved algorithms have been suggested to improve performance of suprathreshold perimetry (165, 166).

Figure 5.14 Adaptive strategy used in automated static perimetry. A: When a constant luminosity is presented throughout a portion of the visual field, true defects near fixation may be missed (falsenegative), whereas more peripheral normal areas may be read as abnormal (false-positive). B: The adaptive strategy minimizes this by changing the stimulus value according to the retinal threshold contour. With full threshold programs, the retinal threshold is crossed by increasing or decreasing the stimulus value (1) and is then crossed a second time with smaller increments of change in luminosity

(2).

Full-Threshold Perimetry

Threshold static perimeters are capable of various testing strategies in addition to suprathreshold screening. The most commonly used programs measure the retinal threshold at 70 to 80 points within the central 24 to 30 degrees. A suprathreshold target is first presented, and the luminosity is then gradually increased or decreased until the patient's threshold is crossed—that is, the target comes in to or goes out of view, respectively. The threshold is then crossed a second time with smaller increments of change in luminosity to refine the threshold determination. Many programs continuously adjust subsequent stimulus values according to prior measurements; for example, the level is increased when testing near a known scotoma on the basis of optimized algorithms. Special programs have been evaluated that automatically increase the density of test locations around defective areas, although the value of this approach has yet to be established (167, 168). Other programs are designed to reduce testing time by adjusting the initial target values according to previous fields by the same patient or by thresholding only locations that are missed with the suprathreshold target. The latter strategy, when compared with full-threshold programs, reduced the testing time by as much as two thirds but missed some defects that were detected with full thresholding (169, 170).

Other Threshold-Testing Algorithms

FASTPAC. Another thresholding strategy to reduc testing time is the FASTPAC program of the Humphrey field analyzer, which estimates thresholding from a single threshold crossing in 3-dB increments, in contrast to the standard double threshold crossing with 4 and 2 dB. This strategy has been evaluated by several investigative teams, most of whom agree that it provides time reduction at some expense of accuracy and reliability (171, 172, 173 and 174).

Swedish Interactive Threshold Algorithm (SITA). In recent years, the relatively new threshold strategy known as SITA has become increasingly popular (175, 176, 177, 178, 179, 180, 181 and 182). This algorithm uses standard 24-2 or 30-2 patterns to assess the visual field on the basis of the probability analysis of the patterns of glaucomatous damage; it is more time efficient than standard threshold strategies. It significantly minimizes test time without significant reduction of data quality. Two versions of SITA are currently available: SITA Standard and SITA Fast. SITA Standard takes approximately half the time to complete, compared with the standard full-threshold program, and SITA Fast takes about half the time of the FASTPAC algorithm. SITA requires significant computer power during the test and is available only on newer Humphrey visual field analyzers.

SITA uses new concepts, such as visual field modeling, that utilizes frequency-of-seeing curves for patients with and without glaucoma. During the SITA test, a computer also produces an information index, which stops the test at the location being examined when threshold reaches a preselected level. The SITA method also makes more individual adjustments to patient response time. After the test is complete, the program makes additional, more precise recalculation of all thresholds measured and produces estimates of false-positive and false-negative response rates (1). One retrospective study found that defects assessed with SITA were often more pronounced, when compared with standard fullthreshold perimetry, but there were essentially no significant differences in quality. Average time reduction by SITA Standard depended on the severity of glaucomatous stage. No significant time difference was found for advanced glaucoma, whereas normal fields using SITA were performed in half the time of full-threshold strategy. The reduction of test time reduces the fatigue factor and permits more frequent visual field examinations and thus a better detection of early glaucoma or progressing visual field damage (183).

Tendency-Oriented Perimetry (TOP). TOP is another fast strategy algorithm available on new Octopus perimeters (184, 185). It also uses a computational approach to estimate threshold values by extrapolating information from surrounding test points. One study compared SITA Fast and TOP P.106

technologies, and found that the mean testing time for the TOP strategy was slightly more than 2.5 minutes, compared with approximately 4 minutes for SITA Fast (186). However, another report