patient and the location of the test point, the status of surrounding points can help determine whether a small deviation is significant.

A mildly depressed point has a greater likelihood of being pathologic if its neighboring points are also depressed.36,37

Fluctuation also increases locally in areas of reduced sensitivity.13 There are several possible explanations for this. It is well known that an area of inconsistency often precedes a permanent depression with Goldmann perimetry. Inconsistent responses on Goldmann perimetry appear as increased fluctuation on computerized automated perimetry. Another explanation is that the nature of the test for fluctuation – repeating the test twice during the examination – means that different areas of the field may be tested because of a small fixation shift. In an area of pathology, the second test can examine a slightly different area that legitimately has different ­sensitivity. The machine compares the first and second tests and indicates the difference between them as SF.

Long-term fluctuation

Long-term fluctuation is that which occurs between two separate visual field tests.This is discussed further in the section on recognition of change, p. 122–125.

Global indexes

Global indexes, which reflect the results of the visual field examination, are mathematic summaries of the actual sensitivity data produced by the examination (Figs. 10-15 and 10-16).

Mean sensitivity

Mean sensitivity is the average of the patient’s responses for all of the points tested.

Mean deviation or defect

Mean deviation or defect (MD) is the measurement of how the mean of the patient’s responses varies from the mean of the responses of a series of normal patients of similar age under similar testing conditions. It is a statement of the generalized depression of the visual field and is useful in recognizing early diffuse visual field loss in glaucoma. Although the MD can be decreased with local defects as well as general ones, there is no way to distinguish large local defects from generalized ones just from the MD (Fig. 10-16A.)

Standard deviation or variance

The standard deviation of the mean of the patient’s responses is the same as the square root of the variance. The Humphrey perimeter analysis program reports standard deviation (pattern standard deviation, PSD), whereas Octopus reports variance (loss variance, LV). Each is a measurement of the variation in responses across the visual field. Normal patients have a small standard deviation, indicating a ‘smooth’ surface to the hill of vision. A high standard deviation or variance indicates an irregular surface to the

hill of vision and may be indicative of localized visual field damage.33,40,42 These indexes can be corrected by SF and then are

labeled as corrected (i.e., corrected pattern standard deviation, CPSD or corrected loss variance, CLV). When the indexes are corrected, they become more sensitive to recognizing true localized defects in the visual field because the variability caused by SF is removed.

Graphic plots

One of the greatest values of computerized perimetry lies in the ability of the computer to analyze the numeric data and present

Fig. 10-15  Detail of visual field indexes from Humphrey field analyses. MD, mean deviation; PSD, pattern standard deviation; SF, short-term fluctuation; CPSD, corrected pattern standard deviation. P values indicate the likelihood that values are normal. These values are severely disturbed.

them graphically for easier comprehension. The grey-scale printout of the Octopus was the first of these. Printouts that show variation from normal are widely available today (Fig. 10-17). The Humphrey STATPAC analysis printout includes probability maps that allow quick assessment of the likelihood that a response is disturbed. Patterns of disturbed points are easily detected. These printouts can also be adjusted for generalized depression so that scotomata are more obvious.This latter function is especially useful for follow-up of patients with glaucoma and other causes of generalized depression such as constricted pupils or cataracts (Figs 10-18 and 10-19).

Area of the visual field to be tested

Most computerized perimeters measure the threshold sensitivity of 56–72 locations in the central 24–30°. Flammer and co-workers46 and Jenni and co-workers45 developed the G-1 program to focus more attention on the areas of particular interest in glaucoma (see Fig. 10-10).

Long-term analysis

By selecting a series of similar tests, computerized perimeters can analyze the results over time to evaluate whether the visual field is changing. This may be accomplished by linear regression or t-test techniques.47 Even with these methods, however, several visual fields of good reliability from the same eye are needed to

analyze change well. Trying to recognize change from fewer fields is fraught with hazard and requires a large amount of change.48,49

This is discussed further in the section on recognition of change, pp. 122–125.

Determination of normal visual field

Evaluation of the visual field in glaucoma patients or glaucoma suspects attempts to answer two important questions: is the visual field abnormal (detection), and has it worsened (progression)?

There are three basic ways by which the physician can determine whether the field is abnormal: (1) by recognizing deviation from normal values; (2) by recognizing variation between two eyes of the same patient, and (3) by establishing significant variation within the given field. Several of the computerized machines have analysis

programs that will indicate abnormalities by graphic plots or by a written phase.36,37,50,51