Humphrey Visual Field Testing (Automated Static Perimetry)

Automated static threshold perimetry — most commonly performed on the Humphrey Field Analyzer — quantifies visual sensitivity at many individual points across the visual field, producing a reproducible, quantitative map used to detect, characterize, and monitor visual field loss from glaucoma, neuro-ophthalmic disease, and other optic pathway disorders. It is the field-testing standard against which confrontation testing and other screening methods are compared.

Before testing, the correct near-refractive lens is confirmed in the trial lens holder for the patient's age and correction, the tested eye is centered and in focus, and the untested eye is patched. The appropriate test strategy and pattern — commonly 24-2 or 30-2 SITA Standard or SITA Fast — is selected based on the clinical question, with a narrower glaucoma-focused pattern used for glaucoma screening and monitoring and a wider field used when neurological disease is suspected. The patient is instructed to fixate steadily on the central target throughout and press the response button whenever a light stimulus is perceived, even if only dimly seen, without deliberately searching for stimuli. The machine presents stimuli of varying intensity at pre-set locations using an adaptive thresholding algorithm to estimate the sensitivity threshold at each point, while the examiner monitors for patient fatigue, fixation loss, and automatically tracked false-response indicators.

At test completion, the reliability indices — fixation losses, false positive and false negative rates — must be reviewed before the sensitivity map itself is interpreted, since an unreliable test cannot be meaningfully interpreted. A normal test shows full, reproducible sensitivity across the field corresponding to age-adjusted normative values with reliability indices within acceptable limits. Glaucomatous defects characteristically include a nasal step, arcuate (Bjerrum) scotoma, and paracentral defects, respecting the horizontal meridian due to retinal nerve fiber layer anatomy. Neuro-ophthalmic defects instead respect the vertical meridian, as with a homonymous hemianopia or quadrantanopia from retrochiasmal disease, or show bitemporal patterns from chiasmal compression such as a pituitary lesion. An unreliable test, with high fixation loss or false response rates, or a scattered "clover-leaf" pattern from inattentive or inconsistent responses, should prompt repeat testing rather than clinical interpretation.

Common mistakes include using outdated or incorrect refractive correction, which artificially depresses sensitivity across the field (a lens rim or edge artifact can also cause a spurious peripheral defect); failing to monitor for and account for patient fatigue, particularly with longer strategies or first-time patients; over-interpreting a single field without considering the learning effect common on a patient's first one or two tests, or without comparing to prior fields to assess for a genuine trend versus test-retest variability; and ignoring poor reliability indices when interpreting a printout.

Deep learning-based analysis of serial visual fields has shown strong discrimination for detecting true glaucomatous progression compared to conventional statistical criteria such as Anderson's criteria, the Glaucoma Hemifield Test, and pattern standard deviation-based analysis in early studies, achieving a notably higher area under the curve than these conventional approaches.¹ This remains an emerging approach, however, requiring further external validation before routine adoption in place of established statistical criteria. Testing conditions also matter: non-standard conditions such as yellow-tinted lens filters, sometimes used for glare or comfort, can measurably alter standard automated perimetry indices, underscoring the importance of testing under the standard conditions the normative database was built on when tracking change over time.²