Abstract
Purpose: :
To compare the Glaucoma Probability Score GPS) with the Moorfields Classification (MFC) for discriminating between glaucoma and healthy eyes using HRT version 3.0 software.
Methods: :
104 glaucoma eyes with repeatable standard automated perimetry damage and 93 normal eyes were included from the longitudinal Diagnostic Innovations in Glaucoma Study (DIGS). The severity of glaucomatous visual field defects ranged from mostly early to some severe damage (average PSD was 5.8 + 5.1 dB). HRTII software version 3.0 includes the calculation of the Glaucoma Probability Score (GPS). The GPS utilizes 2 measures of peripapillary retinal nerve fiber layer shape (horizontal and vertical retinal nerve fiber layer curvature) and 3 measures of optic nerve head shape (cup depth, rim steepness and cup size) as input into a relevance vector machine learning classifier that estimates a probability of having glaucoma. No contour line or reference plane is used in the GPS calculation. The Moorfields Classification (MFC) compares measured rim area to predicted rim area adjusted for disc size to categorize eyes as outside normal limits, borderline or within normal limits. MFC relies on a contour line and reference plane for its measurements. Sensitivity, specificity and likelihood ratios were compared for both global and regional results.
Results: :
The mean (95% CI) GPS was significantly higher in glaucoma eyes (70.9% (66.3,75.5%)) than normal eyes (27.2% (21.4%, 31.1%)). Using manufacturers' suggested cut–offs for GPS Global classification (>64% as outside normal limits), the sensitivity and specificity (95% CI) were 72.1% (62.8%, 79.8%) and 88.2% (80.0%, 93.3%), respectively. The sensitivity and specificity (95% CI) of the MFC result were 67.3% (57.8%, 76.5%), and 91.4% (83.9%, 95.6%), respectively. Positive likelihood ratios for regional GPS and MFC outside normal limit results ranged from 6.0 to 14.7, and 7.8 to 45.6, respectively.
Conclusions: :
GPS tended to have higher sensitivities, and somewhat lower specificities and positive likelihood ratios than MFC. These results suggest that GPS can differentiate between glaucoma and healthy eyes with relatively good sensitivity and specificity.
Keywords: imaging/image analysis: clinical