Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 7 (6), 36
eCollection

Improving Visual Field Examination of the Macula Using Structural Information

Affiliations

Improving Visual Field Examination of the Macula Using Structural Information

Giovanni Montesano et al. Transl Vis Sci Technol.

Abstract

Purpose: To investigate a novel approach for structure-function modeling in glaucoma to improve visual field testing in the macula.

Methods: We acquired data from the macular region in 20 healthy eyes and 31 with central glaucomatous damage. Optical coherence tomography (OCT) scans were used to estimate the local macular ganglion cell density. Perimetry was performed with a fundus-tracking device using a 10-2 grid. OCT scans were matched to the retinal image from the fundus perimeter to accurately map the tested locations onto the structural damage. Binary responses from the subjects to all presented stimuli were used to calculate the structure-function model used to generate prior distributions for a ZEST (Zippy Estimation by Sequential Testing) Bayesian strategy. We used simulations based on structural and functional data acquired from an independent dataset of 20 glaucoma patients to compare the performance of this new strategy, structural macular ZEST (MacS-ZEST), with a standard ZEST.

Results: Compared to the standard ZEST, MacS-ZEST reduced the number of presentations by 13% in reliable simulated subjects and 14% with higher rates (≥20%) of false positive or false negative errors. Reduction in mean absolute error was not present for reliable subjects but was gradually more important with unreliable responses (≥10% at 30% error rate).

Conclusions: Binary responses can be modeled to incorporate detailed structural information from macular OCT into visual field testing, improving overall speed and accuracy in poor responders.

Translational relevance: Structural information can improve speed and reliability for macular testing in glaucoma practice.

Keywords: ganglion cells; glaucoma; optical coherence tomography; perimetry; visual field.

Figures

Figure 1
Figure 1
Matching of fundus images from the Compass perimeter and the Spectralis SD-OCT. (A) Exemplar fundus image of a glaucoma patient with the locations of the 10-2 grid superimposed, as recorded by the tracking system. The red outline indicates the fundus area that was matched with the image from the Spectralis. (B) Matched image from the Spectralis distorted using a projective transformation and superimposed to the fundus image from Compass. The transformation establishes a two-way relationship between the structural map produced by the OCT and the functional map produced by the perimeter. As such, it can be used to map the tested locations from Compass onto the macular OCT map and vice versa.
Figure 2
Figure 2
Quantification of the local macular damage. (A) Exemplar estimated GCC map derived from the segmented GCL in a macular OCT scan. The black squared regions are centered on each one of the 68 locations in the 10-2 grid and have the same area as a G-III stimulus. (B) The same regions displaced and distorted using the model proposed by Drasdo et al. The anisotropic distortion is a consequence of the shift of the grid center from the anatomical fovea.
Figure 3
Figure 3
Example of POS curves for different values of local GCD, as predicted by the structural OCT measurement. The logistic model predicts steeper curves for higher GCD values, centered on higher thresholds. Conversely, for lower GCD values, the POS curves were shallower and centered on lower thresholds. For this example, different from the actual structure-function model, the curves were calculated without taking the eccentricity and age into account.
Figure 4
Figure 4
Example of starting prior distributions for the two Bayesian strategies used in the simulations. (A) Prior distributions of four parafoveal locations with the standard ZEST. The four curves are identical. (B) Prior distributions for the same points derived from the structure-function model. For the damaged location, the abnormal component is weighted more, with the mode of the “normal” portion of the curve centered on lower sensitivities.
Figure 5
Figure 5
Accuracy of predictions of the 50% thresholds from the structure-function model for subjects in group 2. On the horizontal axis, the real threshold values estimated by averaging three perimetric tests. On the vertical axis, the prediction from the structure-function model. The purple dots represent each single location, while the black line represents the ideal perfect equivalence. Due to the bottom floor effect, the thresholds are largely overestimated below 20 dB.
Figure 6
Figure 6
Results of simulations (500) with reliable responses (top panels), 20% FP errors (middle panel), and 20% FN errors (bottom panel) for the MacS-ZEST (left panels) and the ZEST (right panels). The graphs report the error in decibels (vertical axis) at different input thresholds (horizontal axis). The black lines represent the mean error, while the shaded areas represent the 95% quantiles (enclosed in the blue lines) and the 99% quantiles (enclosed in the red lines). No improvement in precision was evident in reliable subjects, while an important shrinkage of the 95% and 99% quantiles was evident with high FP and FN errors.
Figure 7
Figure 7
Average number of presentations per location (solid line) at different sensitivities with simulated reliable responses (top panels), 20% FP errors (middle panel), and 20% FN errors (bottom panel) for the MacS-ZEST (red) and the ZEST (green). The graphs report the number of presentations (vertical axis) at different input thresholds (horizontal axis). The vertical bars represent the SDs at each simulated input threshold. A reduction in presentations was obtained for thresholds higher than 10 dB.
Figure 8
Figure 8
The left panel shows the mean difference per test in MAE between MacS-ZEST and ZEST for reliable subjects (dark gray) and for increasing levels of FP (red) and FN (blue) errors. The middle panel shows the same difference as percentage reduction. The right panel reports the percentage reduction per test in the number of presentations obtained with MacS-ZEST. The vertical bars represent the 95% CI for 500 simulations. Each test included eight locations (the small grid used for group 2). CI, confidence interval.
Figure 9
Figure 9
Average number of presentations for full 10-2 fields (68 locations) for both strategies for reliable responses (top), 20% FP errors (middle), and 20% FN errors (bottom). The difference in speed between the two strategies was negligible for fields with a mean sensitivity below 10 dB. The vertical bars indicate the SD around the mean.

Similar articles

See all similar articles

References

    1. Glen FC, Baker H, Crabb DP. A qualitative investigation into patients' views on visual field testing for glaucoma monitoring. BMJ Open. 2014;4:e003996. - PMC - PubMed
    1. Turpin A, McKendrick AM. What reduction in standard automated perimetry variability would improve the detection of visual field progression? Invest Ophthalmol Vis Sci. 2011;52(6):3237–3245. - PubMed
    1. Artes PH, Iwase A, Ohno Y, et al. Properties of perimetric threshold estimates from Full Threshold, SITA Standard, and SITA Fast strategies. Invest Ophthalmol Vis Sci. 2002;43:2654–2659. - PubMed
    1. Bengtsson B, Heijl A. Evaluation of a new perimetric threshold strategy, SITA, in patients with manifest and suspect glaucoma. Acta Ophthalmol Scand. 1998;76:268–272. - PubMed
    1. Watson AB, Pelli DG. QUEST: a Bayesian adaptive psychometric method. Percept Psychophys. 1983;33:113–120. - PubMed

LinkOut - more resources

Feedback