Kinetic perimetry

Kinetic perimetry is typically performed manually by confrontation, on a tangent screen, or with a Goldmann perimeter. In kinetic perimetry, the stimulus usually is presented in the non-seeing periphery and moved at approximately 2° per second toward fixation until the patient first perceives it. The stimulus is subsequently moved to another meridian in the periphery out of view and advanced toward fixation again until the patient sees it. By repeating these maneuvers at approximately 15° intervals around 360° of the visual field, the examiner defines a series of points that can be connected to describe an isopter corresponding to the stimulus used (see Fig. 8-1). By decreasing or increasing the size or brightness of the stimulus, a smaller or larger isopter will be outlined. If the stimulus is presented into randomly selected areas of the visual field, the isopters will be slightly constricted and irregular compared with sequentially presented stimuli. Reproducibility may be greater with sequentially presented stimuli.6

After initial detection, a scotoma can be defined more precisely with kinetic perimetry by placing the stimulus in the scotoma and moving the stimulus outward until it is perceived. This process is repeated in various directions until all edges of the scotoma have been defined. If the edges of the scotoma are sloping (the change from normal to abnormal regions within the field is gradual), a brighter stimulus will define a smaller scotoma, and a dimmer stimulus will define a larger scotoma. If the margins of the defect are steep, changing the stimulus size or intensity will affect the size of the scotoma only slightly.

Static perimetry

In static perimetry, the test stimulus size usually remains constant throughout the test. For computerized full-threshold testing, each point in the visual field is evaluated by positioning the stimulus at a test point and varying the intensity until the threshold for that particular retinal location is defined (Fig. 8-3).This process is repeated

Fig. 8-4  Humphrey 700 series perimeter.

until all of the positions of the retina to be measured have been tested.

The more retinal positions tested, the more defects will be found and quantified. There is, however, a point of diminishing returns, at approximately 80 locations, wherein patient fatigue seriously reduces the accuracy and consistency of responses.7,8 Most computerized perimeters (Figs 8-4 and 8-5) use static visual field testing techniques for their standard tests.

Alternatives and modifications to standard full-threshold testing of each retinal position have been devised to reduce the number of patient responses required without reducing the amount of information obtained at each testing session.9–12 Such alternatives include threshold-related testing and zone testing, as well as

Fig. 8-5  Octopus 123 perimeter.

algorithms that use less precise bracketing to estimate the threshold.These methods generally produce results that are similar to, but

can be somewhat more variable than, standard threshold determining strategies.13–16 The most widely used modification of stand-

ard repeat-bracketing threshold testing is the Swedish Interactive Testing Algorithm (SITA) program used on the Humphrey Field Analyzer, which adjusts the starting and ending points of the bracketing procedure during the examination based on the patient’s responses.17,18 This is done in a fashion that reduces redundancy, decreases testing time, and increases accuracy and patient accept-

ance without compromising the sensitivity and specificity of the test (see also p. 95).19,20

Threshold-Related Testing

The ‘normal’ state of the visual field is a statistically determined

figure obtained from the testing of many normal individuals of different ages, genders, etc.21,22 Each retinal location has a statisti-

cally determined ‘normal’ sensitivity range that can be expressed in decibels of stimulus intensity related to stimulus size, background intensity, and patient age.23 This sensitivity is not constant from patient to patient, or even within the same patient from test to test. Therefore, for a particular retinal location to have a strong possibility of being abnormal, its sensitivity should be reduced from normal by roughly two standard deviations of the mean of normal, or approximately 4 or 5 dB on average.This is conveniently expressed as twice the average short-term fluctuation (SF). It is similar to the traditional rule of thumb that suggests that significant measurement deflections are at least twice the baseline noise level. The average short-term fluctuation is lower at or near fixation than it is in the periphery, so a deviation of 4 or 5 dB centrally has a greater chance of representing a reproducible change in sensitivity than does a defect of similar depth in the periphery. A 4 dB depression in an area of the field that has a SF of 1.2 dB is more likely to represent pathology than a 10 dB depression in an area than has a SF of 8 dB, which can be the case at 24° or more from fixation.

In threshold-related testing (Fig. 8-6), if the patient is presented with a stimulus that is roughly 4 dB brighter than the expected normal level for that retinal position and the patient sees it, the location is considered normal, and the stimulus is moved to the next position without measuring the threshold of the location precisely.

Fig. 8-6  Threshold-related testing. (A) A single stimulus, usually 4 or 5 dB brighter than the anticipated threshold, is exposed across the visual field. If the patient sees it, that part of the field is considered normal. (B) Defect 1 will be detected by the technique, but defect 2 will be missed.

The disadvantage of this technique is that it only finds defects equal to or greater in depth than the suprathreshold stimulus used. This technique also provides no information regarding subtle variation in the contour of the field, which is important in recognizing early changes from normal.24 The rapidity of testing normal areas using the technique, however, allows a larger area of the retina to be examined. If defects are detected, they can be quantified with the full-thresholding strategy.

Zone Testing

Zone testing uses three levels, or zones, of stimulus intensity to locate and then quantify defects. The first zone is a suprathreshold stimulus 4 or 5 dB brighter than the anticipated normal threshold, as described in the section on threshold-related testing, above. If the patient sees this stimulus, the response is recorded as normal. If the patient fails to see the initial stimulus, a maximally bright stimulus is shown. If the patient sees this stimulus (but failed to see the initial, relatively dim, stimulus), the machine indicates a relative defect. If the patient fails to see either stimulus, the machine records an absolute defect. Responses can thus be grouped in three zones – normal, relative defect, or absolute defect. There are multiple variations on this theme that allow a greater number of zones to be defined, or for zones to be defined at different levels. The obvious disadvantage of this technique is that subsequent testing can only recognize major change because the difference between the test stimuli is great.The advantage is that it is fast.

Screening Tests

Screening tests for visual field defects are available by manual per-

imetry and with most computerized perimeters. Unfortunately, they only detect rather large changes in the visual field.25–27 Most

screening programs use a technique that recognizes defects that are greater than 4 or 5 dB below an expected level. As such, they may not detect early glaucomatous defects. Also, if a defect is found, the