|Year : 2019 | Volume
| Issue : 3 | Page : 220-230
Interpreting HFA single field reports
Panda Smita1, George Ronnie1, Murali Ariga2
1 Smt. Jadhavbai Nathamal Singhvee Glaucoma Services, Sankara Nethralaya, Chennai, Tamil Nadu, India
2 Sundaram Medical Foundation, Chennai, Tamil Nadu, India
|Date of Submission||24-Jul-2019|
|Date of Acceptance||25-Jul-2019|
|Date of Web Publication||11-Nov-2019|
Dr. Panda Smita
Smt. Jadhavbai Nathamal Singhvee Glaucoma Services, Sankara Nethralaya, Chennai, Tamil Nadu
Source of Support: None, Conflict of Interest: None
Autoperimetry is a very important investigation for glaucoma diagnosis and management. This article deals with the basics and terminologies used in humphrey visual field analysis and deals with the interpretation of humphrey visual field analysis printout.
Keywords: Humphrey field analyzer, single field, visual fields
|How to cite this article:|
Smita P, Ronnie G, Ariga M. Interpreting HFA single field reports. TNOA J Ophthalmic Sci Res 2019;57:220-30
| Introduction|| |
Visual field is the area of the external environment of an observer, wherein the observer can detect a visual stimulus with the eye fixated. As famously described by Traquair, it is the hill of vision surrounded by sea of darkness which extends 60° nasally, 90° temporally, 50° superiorly, and 70° inferiorly.
A perimeter can be based on kinetic strategy or static strategy. A kinetic perimeter works by presenting various stimuli with varying sizes/intensities which are moved from nonseeing points to seeing points. For a stimulus of particular strength and size, all the seeing points are joined to form its isopter. In static perimetry, a particular point is tested by varying stimuli intensity. Hence, in the kinetic perimeter, the stimuli move, and in the static perimeter, the stimuli do not. Goldman kinetic perimeter is an example of kinetic perimeter, and Humphrey visual field analyzer (HFA) and octopus are the most commonly used static perimeters. Kinetic programs are also available in Humphrey and octopus but not used commonly.
| Humphrey Visual Field Analyzer|| |
The analyzer projects a series of white light stimuli of varying intensities (brightness), throughout, on a uniformly illuminated bowl. The patient is given a hand-held button that is to be pressed to indicate that the light is seen.
A few basic principles of perimetry help in understanding why the test is performed in this way and in interpreting test results.
While the unit of luminosity used in HFA is apostilb (asb), light intensity to which the patient responds positively is recorded in “decibels” (dB). dB and asbs are inversely related with low dB values corresponding to brighter stimuli than high dB values. One decibel is described as one-tenth of log unit. If there is ten times increase in the intensity required to stimulate a particular point of retina, retinal sensitivity has decreased by 1 log unit or 10 dB and 100 times increase in required intensity makes a 20-dB reduction.
For each point in fields, the threshold value is determined. It is that intensity of the stimulus, which when presented to a particular point, has probability of being detected 50% of the time. The fovea has the highest threshold that is it can see the least intense stimuli and the sensitivity progressively decreases toward the.
There are numerous testing protocols to select, based on the purpose. The first number denotes the extent of the field measured on the temporal side, from the center of fixation, in degrees. They include:
- 10-2 measures 10° temporally and nasally and tests 68 points
- 24-2 measures 24° temporally and 30° nasally and tests 54 points
- 30-2 measures 30° temporally and nasally and tests 76 points.
When testing the central 30° and central 24° with the Humphrey perimeter, the grid points are spaced 6° apart. The central horizontal points are distributed at equal distance from the horizontal axis that is 3° above and below – this program is known as 30-2 and 24-2, respectively. The advantage of using the central 30-2 or 24-2 programs lies in the spacing of the central horizontal points at 3° from the center of the field, with greater sensitivity to changes across the horizontal retinal raphe.
This machine has evolved from HFA1 to HFA3. HFA3 provides a streamlined and faster workflow with the help of new features incorporated into it, such as:
- Reduce visual field-testing time with New Swedish interactive thresholding algorithm (SITA) Faster
- Mixed guided progression analysis: Inter-mixing of SITA Standard, SITA Fast, and SITA Faster
- Newer SITA 24-2 algorithm with more central points.
All Humphrey strategies other than suprathreshold strategy start testing at a single location in each quadrant of the visual field. If a stimulus is seen, subsequent stimuli at that location are dimmed one step at a time – usually by 3 or 4 dB – until they are no longer seen. On the other hand, if the initial stimulus is not seen, then subsequent stimulus presentations are made brighter in steps until the patient presses the response button. Some strategies such as older full-threshold strategy repeat this process for confirmation of the finding, either using the same step size, or perhaps a smaller step, such as 2 dB.
- Suprathreshold: Stimulus presented is brighter than the threshold value required at that point. In a suprathreshold strategy, a single bright suprathreshold stimulus is presented across the entire hill of vision. The response is simply recorded as seen or not seen. It is a rapid strategy and can be used for screening purpose
- Full threshold: It is older strategy which has been abandoned since inception of SITA. It is seldom used because of long duration of about 15–20 min.
Swedish interactive thresholding algorithm strategy
SITA strategies make much more precise use of information than the earlier full-threshold and FastPac™ strategies. SITA strategies gain efficiency by ceasing to present stimuli at a given location when predetermined statistical levels of testing certainty are reached, based upon the consistency of patient responses. This method allows test time to be shortened when reliably consistent responses are given and extended when there still is uncertainty. The overall effect is reduced testing time without loss of diagnostic information. SITA Fast and SITA Faster are quicker methods of testing. They produce similar results compared to SITA Standard; however, repeatability is not as good and the tests are slightly less sensitive.
| The technician and the Patient|| |
It is particularly important to tell the patient what to expect during the test. Technicians who have undergone visual field testing themselves will be better prepared to brief patients. The technicians should explain the patient's task, show him or her what the stimulus will look like, where it might appear, how long the test will last, when blinks are allowed, how to sit, and how to pause the test. The patient should understand that more than half of the stimuli shown in a threshold test will be too dim to be seen and that the stimuli that are seen are likely to be barely visible.
When a perimetric test is needed, a 30-2 or 24-2 size III, white SITA Standard threshold test is the best choice in most cases. In cases of advanced glaucoma, many or most points may show absolute defects with size III stimuli, jeopardizing perimetric follow-up with a central 24-2 or 30-2 test. One can then switch to the central 10-2 test or size V stimulus, which is four times the diameter of size III. Testing with size V stimuli will result in sensitivity levels that are 5–10 dB higher than those found using size III, often extending the available sensitivity range and making it possible to follow such patients. It should be noted that if the size V stimulus is used, one no longer has access to several of the analytical follow-up tools available for the standard size III tests.
For easy interpretation of the visual field printout, we divide the printout into eight zones. We describe the relevance of each zone on interpretation.
| Zone-Wise Interpretation|| |
It tells us which test has been performed and on whom. In a busy clinic, checking the name is important because there are chances that wrong fields could get uploaded or the patient has taken the wrong test report with himself/herself. One needs to check if a technician has done the test strategy you ordered. One can also look at the type of target used, the strategy, and whether the fixation target was central or a diamond. A patient with central scotoma may not be able to fixate with central target and give high errors and false results. Correct entry of patient's birth date is important because the patient's threshold will be compared to the normal individuals of the same age. One has to make sure that the refractive correction given is appropriate as an uncorrected refractive error can result in depressed fields. Check the pupil size. Pupil sizes <2 mm can give falsely constricted fields, and further, two fields can be compared with each other provided pupil size was similar. Visual acuity also is given in these data.
When we look at patient's visual acuity, we compare it with foveal threshold. Visual acuity is best at the fovea. If the visual acuity is good but the foveal threshold is low, then there may be early damage to the fovea. On the other hand, if the foveal threshold is good and the visual acuity is low, the patient may have been using incorrect glasses and needs refraction.
Zone 2 also provides the reliability criteria: fixation loss (FL), false negative (FN), and false positive (FP).
Three indices of reliability are presented to help identify tests that are unlikely to contain reliable diagnostic information: the FP rate, the FN rate, and the FL rate.
The FP error score measures the tendency of a patient to press the response button even when no stimulus has been seen. FP is determined in two different ways, depending upon which testing strategy is being used. With the SITA strategies, FP rates are determined as a function of patient responses – presses of the response button – made during times that patient responses are not expected. The older thresholding methods calculated FP rates by recordings patient responses made when the instruments made all the normal testing motions and sounds but did not actually display the stimulus. In either case, FP response rates are presented as a fraction or percentage of incorrect patient responses. FP rates exceeding 15% may indicate compromised test results.
Trigger-happy patients – those who press the button even when no stimulus is seen – usually produce misleading visual field tests. Such results are usually indicated by an elevated FP score, but they are also frequently identified with the “abnormally high sensitivity” message on the glaucoma hemifield test (GHT) and by the presence of white areas on the gray scale, indicating impossibly high thresholds. Fields that show much larger defects in the pattern deviation (PD) plots than in the total probability plots may be the result of high numbers of FP answers.
The FN rate measures the tendency of a patient to fail to press the response button even when a distinctly visible stimulus has been presented. FN stimuli are presented only at test point locations where the threshold sensitivity has already been measured and are 9 dB (8 times) brighter than threshold.
Interpretation of FN rates is complicated by the presence of visual field loss. High FN response rates from glaucoma patients seldom have anything to do with the patient's test-taking performance but, instead, usually reflect the decreased reproducibility of glaucomatous visual fields. In general, little may be concluded from the presence of elevated FN rates in abnormal visual fields, and the value of FN rates must be viewed as limited.
Fixation loss rate
How steadily the patient gazes at the fixation stimulus is estimated by the FL score or, in some instruments, by examining the gaze tracking record. FL rates are estimated by the Heijl–Krakau method in which stimuli are periodically presented at the presumed location of the patient's blind spot. Positive patient responses to such stimuli indicate that the patient is not looking straight ahead. FL rates exceeding 20% may indicate compromised test results although high FL levels are frequently seen artifactually when the blind spot has not been properly located and with trigger-happy patients.
The gaze tracker available in most newer models of the Humphrey perimeter has a precision of 1°–2° and records deviations from correct fixation as upward deflections on the record. Full-scale spikes indicate deviations of 10° or more. Even highly experienced perimetry patients show occasional gaze errors, and such fixation patterns are generally associated with reliable test results. Fixation records that show excessive periods of poor fixation interspersed with times of correct fixation may be associated with test results with reduced reliability. Excessive periods of poor fixation tend to be associated with test results in which field defects have indistinct edges and appear to be shallower than they really are.
This is a graphic representation of the recorded threshold sensitivities in the numeric scale. Regions of decreased sensitivities are displayed in darker tones. We do not make a diagnosis based on the gray scale. It shows white scotomas when there is gross FP error. It occurs in trigger-happy patients as discussed earlier, and one can see cloverleaf pattern in patient with high FN error. It is also useful to demonstrate the patient the kind of visual field defect they have.
This is the total deviation (TD) plot.
Total deviation probability plots
TD probability plots indicate all test locations that are outside normal limits, whether because of a general depression of the whole visual field or because of localized loss. Threshold sensitivity is compared with the age-corrected normal values at each test point to produce the TD dB plot. Negative values indicate sensitivities that are below the median age-corrected sensitivity, and positive values indicate higher than normal sensitivities. The normal range of sensitivity is larger in the periphery than at the center of the field and also larger superiorly than inferiorly. Thus, a deviation of 5 dB from age normal may be quite significant at the center of the field but totally within the normal range of variability in the periphery of the test area.
The significance of these deviations from normal is indicated in the associated TD probability plot, in which sensitivities that are worse than those found in 5%, 2%, 1%, and 0.5% of normal subjects of the same age as the patient are highlighted with appropriate symbols. A 2% symbol, for instance, indicates that fewer than 2% of normal patients have sensitivity that low or lower. A key showing the meaning of the symbols is given near the bottom of the printout.
The machine adjusts for overall depression of the visual field, due to cataract or some other reason, and gives this zone. It adjusts for any overall sinking of the hill of vision, and in the PD plot, it draws our attention to any localized scotomas that may have been hidden inside this depressed visual field. Thresholds across the entire field are recalculated usually to adjust for generalized visual field depression because of a cataract, corneal opacity, or miosis – this generalized loss can mask a localized glaucomatous defect. Once the field is recalculated after adjusting for this depression, a localized scotoma can be detected. The PD plot is also provided as a numerical plot as well as a probability plot. We generally look at the probability plot. We try to look for abnormal points in a cluster. The low probability symbols do not matter as much as abnormal points in a cluster in an expected area. The probability plots ignore normal variations and highlight significant but subtle defects that might otherwise escape notice.
Early visual field defects show up earlier in the probability map than in gray scale printouts. Furthermore, STATPAC's probability plots help de-emphasize common artifactual patterns, such as eyelid-induced depressions of sensitivity in the superior part of the field, that are often overemphasized on the gray scale. Note that the correction for homogenous depression used in the PD plot is based upon the sensitivities at the best points in the TD plot; thus, if visual field loss is so far advanced that even the best points are almost blind, then the PD plot will be unable to highlight localized loss. Such situations are obvious even when looking at the gray scale, however, and should not lead to missed diagnoses.
Comparison of total and pattern deviation plots
It is useful to compare the TD and PD plots.
- Uniformly depressed TD plot and normal looking PD plot: cataract or any generalized depression of field
- Both plots look more or less the same: little or no generalized loss
- Normal TD and abnormal looking PD: trigger-happy patient.
The global indices summarize the field. The indices available in SITA strategies are as follows:
It tells how much the total field departs from normal. This is the average deviation from the normative data at all the tested points. It has a negative (−) sign. A small localized defect will show a small mean deviation (MD), whereas a generalized or an advanced defect will show a high MD. The value does not differentiate a generalized and a localized field loss. Positive MD is seen in hypersensitive or trigger-happy patient and normal patients having sensitivity values higher than those in normative database.
Pattern standard deviation
This index gives an idea about the resemblance of the patients' field to the shape of hill of vision. It has a positive sign. Low pattern standard deviation (PSD) indicates a normal shape of the hill, whereas a high value indicates a disturbed shape of the hill. Localized defect will give a high PSD, whereas a generalized defect will give a low PSD.
Visual field index is a new index which expresses the visual field status as a percent of a normal age-adjusted visual field. It is derived from the PSD and MD. Greater weight is given to points closer to fixation. The visual field index (VFI) ranges from 0 to 100, with 100 being field with all normal threshold values and 0 a completely extinguished field.
Interpretation of the global indices
All the indices should be considered together for interpretation and are inter-related. MD is used for determination of the stage of glaucoma damage. PSD is important for the diagnosis of early glaucoma.
- Cataract alone will show a high MD but a normal PSD
- Early glaucoma without cataract will show close to zero MD but high PSD with a low P value
- In cataract and glaucoma or in advanced glaucoma, MD and PSD all will be abnormal with a low P value
- PSD value conversely improves as the field defect covers most of the tested area as in advanced field loss. Hence, this indicator loses its value in advanced disease. Such patients can be followed up with the MD.
The GHT is a plain-language classification of threshold test results in the following categories.
- The “outside normal limits” message is displayed whenever sensitivities in one or more of the five zones in the upper half of the field are significantly different (P< 0.01) from the sensitivities measured in the corresponding zones in the lower half of the field
- Fields are labeled as “borderline” whenever zone pairs differ by an amount greater than is seen in most normal patients (P< 0.03), but the difference does not reach the level required for the “outside normal limits” message
- “General depression of sensitivity” or “abnormally high sensitivity” whenever even the best test point locations are either so low or so high as to be at levels seen in only 0.05% of normal patients. “General depression of sensitivity” will not be displayed, when sensitivity differences between the superior and inferior hemifields are large enough to result in an “outside normal limits” message
- The “within normal limits” message is presented whenever none of the above significance limits are reached.
Sensitivity differences between the upper and lower hemifields are a hallmark of glaucomatous field loss. Asymmetry in the number of abnormal points in any of the five defined regions between the upper and the lower hemifield is common in glaucomatous fields. The GHT has also been reported to be the single most effective method of visual field analysis, and after over 10 years of use, it enjoys wide acceptance worldwide.
The GHT's zone pattern was optimized for the diagnosis of glaucomatous visual field damage, and even if the GHT analysis identifies most abnormal fields, it was not intended for use with other diseases.
Gaze Tracker-Eye Tracking System (gaze monitor) in SITA tells, by deviation above or below a horizontal line, the exact instances when a patient closes their lids (excluding blink) or makes a saccadic deviation from fixation. Deviation upward indicates that the patient's gaze was not on the fixation target. The magnitude of the deflection indicates the extent of the FL. Large deviations downward indicate a blink while small downward deviations indicate that the computer cannot tell the direction of the patient's gaze. A pattern that resembles a city skyline indicates dubious reliability.
Anderson and Patella's criteria for glaucomatous field defect
These criteria (in relation to a Humphrey printout only) are helpful in the diagnosis of early glaucoma and are as follows:
- Abnormal GHT
- Three or more nonedge points in a location typical for glaucoma on the 30-2 printout (edge points are also included on a 24-2), contiguous and with a P < 5%, of which at least 1 has a P < 1%. On a 30-2, nasal edge points are also included for classification
- PSD OR CSPD (full-threshold strategy) should be abnormal and should have a P < 5%.
These are made with the 30-2 full-threshold printout of the Humphrey in mind that was the standard strategy in use earlier and less frequently today. All these defects should be reproducible and should be demonstrated on two successive tests.
| Glaucomatous Visual Field Loss|| |
Common patterns of glaucomatous field loss correspond directly to the patterns of optic nerve damage typical of the disease. Glaucomatous visual fields show more variability than normal fields, and variable reductions in sensitivity frequently precede definite loss.
| Anatomy and Glaucomatous Visual Field Defects|| |
Glaucomatous field loss is the result of axonal damage at the level of the optic disc and is therefore the functional correlate of neural loss or reduced neural function.
| Retinal Nerve Fiber Layer and Optic Disc Anatomy|| |
Retinal ganglion cell axons follow an arcuate path to the optic nerve. Axons extending from the optic disc toward the temporal retina curve around the macular area. Neurons from the temporal superior and inferior retinal areas do not mix but respect the horizontal temporal raphe. Axons maintain a retinotopic organization in the optic disc in the sense that longer axons are situated in the optic disc periphery close to the scleral canal, while shorter axons from ganglion cells nearer to the optic disc follow a more central course through the optic disc.
| Common Glaucomatous Field Defects and Their Anatomical Correlates|| |
Glaucomatous visual field defects commonly take the form of paracentral scotomas, arcuate scotomas, nasal steps, and contractions of the nasal field. Several different types of defects often occur concurrently in the same field.
Arcuate defect – The bjerrum scotoma
A focal notch at the optic disc that reaches the edge of the disc will lead to the loss most of all retinal nerve fibers in the area corresponding to the notch and therefore to a deep arcuate field defect connected to the blind spot. It usually extends around the point of fixation and ends abruptly at the nasal horizontal meridian corresponding to the temporal raphe to produce what is known as a Bjerrum defect. Focal notches at both poles of the optic nerve can result in a double arcuate defect.
If the notch is partial, that is, if it involves only a portion of the axons in the affected area of the optic disc, it is likely that the involved fibers are of approximately the same length and originate from only a part of the arcuate segment. The resulting visual field defect is a paracentral scotoma. Paracentral scotomas can occur anywhere in the central field, but they are particularly common nasally.
More widespread involvement of fibers in the optic disc will seldom be entirely symmetrical but usually will involve a larger percentage of lost fibers in either the inferior or superior half of the optic disc. As a result, differential light sensitivity in the opposite visual field halves will not be the same. This is likely to manifest itself as an abrupt difference of sensitivity across the nasal horizontal meridian in the visual field – a nasal step. (Nasal step damage in one hemifield can be combined with loss in the other hemifield).
| Characteristics of Glaucomatous Field Loss|| |
Localized and generalized visual field loss
Paracentral and arcuate scotomas as well as nasal defects are examples of localized field loss, that is, defects that have shape. Generalized, homogenous visual field loss, in contrast, is a uniform loss of sensitivity across the whole visual field, resulting in a depression of the hill of vision without any significant change of its shape.
Early glaucomatous field loss
Early glaucomatous field loss may develop very gradually over a period of several years. Local depressions of sensitivity frequently come and go for quite some time before finally resolving into stable and repeatable defects. The narrower normal limits of SITA mean that statistically and clinically significant defects can be identified in probability plots even before they are clearly visible in gray scale representations. This happens regularly in patients who are developing early glaucomatous visual field loss, and it is therefore important to focus on probability plots rather than gray scale representations.
Glaucomatous visual field variability
Visual field variability can be considerable in glaucomatous eyes. Threshold values at individual test point locations frequently vary from test to test of the same eye, even if the tests are administered within a short period. Such increased local fluctuations also typically precede definite glaucomatous field defects. The random test–retest variability of glaucomatous fields depends on test point defect depth, test point location, and overall visual field status. Overall field status is also important; fields with widespread areas of damage show higher variability than fields with smaller defects.
| Before Interpreting a Field Report|| |
There is a learning curve in automated perimetry. A patient might require test several times to give a reliable field. It is difficult to get a first test reliable, but some of the people do well, to begin with; hence, it may not necessarily be ignored. A field defect should be reproducible, to be considered statistically significant. Field defects are always to be correlated clinically with disc and retinal nerve fiber layer changes. A defect that is bizarre and not clinically correlating can be discarded.
Now we shall try to interpret a field based on the explanation given above.
- Take a look at [Figure 2] and move zone-wise. Zone 1 is 24-2 program. First, you should confirm that the given field belongs to the patient you are assessing, and the birth date was entered correctly. The fixation target used was routine size 3. The correction used was appropriate. The pupillary diameter was 5.6 mm. The visual acuity was 6/6 which correlates with the foveal threshold given in Zone 2.
The reliability criteria in Zone 2 are very well within acceptable limit.
At Zone 3, the gray scale shows scotoma in superior hemifield. However, we never should conclude with this.
In Zone 4, the TD plot shows several points scattered in the superior hemifield which are depressed to an extent expected in < 5% of the population.
At Zone 5, the PD plot, where the machine has adjusted for any overall sinking of the hill of vision, we see the depression at same points persisting. Now we look at the global indices in Zone 6. Except for the MD is decreased, all the global indices are normal. The MD has been flagged as abnormal and there is positive pattern deviation value; this again reflects a localized scotoma. At Zone 7, the GHT finding states “outside normal limits.””.
- All the Anderson's criteria are fulfilled
- GHT is outside normal limit
- PSD is abnormal with P <5% (P< 0.5)
- More than three contiguous points are depressed to <0.5%
- Hence, there is a high probability of the patient having glaucomatous field defect. One can confirm the same by looking at the optic nerve head.
Now let us take another example. Look at [Figure 3]
It is 30-2 Humphrey field analysis of the right eye. First going through the patient's data, testing condition and reliability indices all of which seem fine. The gray scale shows scotoma superiorly and infero-nasally. Again emphasizing on not to get biased looking at the gray scale, we move to the TD plot. The TD plot shows scattered depressed point all around. On PD plot, few points persist in inferior nasal quadrant. On Anderson criteria – GHT is outside normal limit, PSD is abnormal. We ignore the edge points; there are three nonedge points depressed <5%, of which two are <1%. Hence, with this, we conclude there is inferonasal scotoma which could be glaucomatous. However, ultimate decision is only made after a thorough clinical examination, to find if the patient shows corresponding neuroretinal rim thinning in supero-temporal area. If it is not so, then one can repeat the fields to see if the scotoma disappears or persists. Usually along learning curve patient might confuse you with their fields and one should always confirm the defect by repeating the field later.
Let us look at another field, [Figure 4]. The gray scale is depressed all over. The TD plot shows several points that are which are depressed to an extent expected in less than 5% of the population. However, the PD plot reveals no abnormality. This is typical presentation of cataract without glaucoma.
Now let us look at some common artifacts seen in field analysis printouts.
- A clover leaf pattern [Figure 5]: The patient has given excessive FN. This pattern occurs because the machine tests the central points in each quadrant first and then peripheral. A patient is usually alert in the beginning and responds to the centrally projected stimuli and misses peripheral stimulus which come later during the test.
- Rim artifact [Figure 6]: It can be seen when higher positive refractive error corrections are given or with misaligned lens. It occurs at edge points always and a sudden drop of retinal sensitivity is seen at the edge points.
- White scotoma [Figure 7]: A trigger-happy patient who has given excessive FP and has given response to stimuli which is of very low intensity and would normally go unperceived by an individual of that age. These fields will show no depression in TD plot but scotomas in PD plot. This is because the PD plot will highlight the actually normal points as scotomas by comparing them to supranormal points.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Thomas R, George R. Interpreting automated perimetry. Indian J Ophthalmol 2001;49:125-40.
] [Full text]
Anderson DR, Patella VM. Automated Perimetry. 2nd
ed. St. Louis: Mosby & Co.; 1999.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5], [Figure 6], [Figure 7]