advertisement

Editors Selection IGR 18-1

Miscellaneous: A statistical model to help detect visual field progression

Comment by Andrew Tatham on:

69493 A Statistical Model to Analyze Clinician Expert Consensus on Glaucoma Progression using Spatially Correlated Visual Field Data, Warren JL; Mwanza JC; Tanna AP et al., Translational vision science & technology, 2016; 5: 14

Find related abstracts

Visual field (VF) progression can be evaluated by direct comparison of VF reports or by using progression software to obtain objective, quantitative data. Progression software performs event- and trend-based analyses, however, these rely either on global indices such as the visual field index (VFI) or on assessing point-by-point change over time, which results in loss of potentially useful information. Clinicians viewing serial VFs can gauge the spatial relationship between potentially progressing points, something that automated software fails to consider but that may be helpful for improved detection of progression. For example, if there is worsening sensitivity at one VF test location, change at a related location may be more likely to be genuine, whereas change at an isolated location may be more likely to represent noise.

This study examines a new statistical method for detecting VF progression that incorporates spatial information. The model was developed using a dataset of VFs from 191 eyes of 97 patients, with an average of 7.4 VFs per eye obtained over a mean follow up period of 2.6 years. Eyes with 'definite' VF progression were identified by expert clinicians viewing the individual VFs and glaucoma progression analysis (GPA) report. The relationship between rates of change in VF at each test location and the probability of experts deciding an eye was progressing was examined to determine if deterioration at a particular location was more or less predictive of being diagnosed as progressing by experts. The model adjusted the contribution of each test point by accounting for sensitivity of surrounding VF locations, with the spatial locations partially determined using Garway-Heath's map relating VF test locations to regions of the optic nerve head.¹

When applied to a validation dataset, the new glaucoma progression model provided better predictions of progression compared to competing models. The analysis also revealed that change in sensitivity in regions related to the temporal ONH had a greater influence on probability of progressing compared to change in the nasal sectors. Further validation is needed but it is possible that such a model could be used in clinical practice to improve progression analysis and promisingly there is also the possibility of incorporating information from imaging devices to further improve accuracy.

• ← Previous
• Next →