Mousavi, Seyedmostafa Sadegh

Background This study aimed to provide methodology to create a facial contour-informed normative 60-4 visual field and to apply the reference field to a glaucoma cohort to derive facial contour-informed 60-4 global indices. Methods Participant recruitment was completed at a tertiary medical centre. 60 eyes from 30 participants with no ocular pathology by clinical exam or optical coherence tomography were recruited for the healthy cohort. For the glaucoma cohort, 86 eyes from 43 patients diagnosed with glaucoma by clinical exam were recruited, with 30-2 visual field mean deviation used to stratify glaucoma severity. Results The median (range) age of the healthy participants was 33 (24–61) years old. A facial contour-informed 60-4 visual field reference field was developed. 43 glaucoma patients were recruited with a resulting median (range) age of 65 (30–84). Three algorithms compensating for facial contour were used for the derivation of 60-4 mean deviation and pattern SD. All mean deviation algorithms correlated with 30-2 mean deviation severity (ρ≤−0.62, p<0.0001) and retinal nerve fibre layer thickness (ρ≥0.36, p<0.05). For pattern SD, consideration of inferior nasal defects allowed correlation with both 30-2 mean deviation severity (ρ≥0.49, p<0.001) and retinal nerve fibre layer thickness (ρ≤−0.47, p<0.01). Conclusion This study provides methodology for deriving a facial contour-informed 60-4 reference field and illustrates the feasibility of applying facial contour information to refine global indices of the 60-4 visual field. Obtained metrics compensated for facial contour and correlated with functional and structural disease metrics.

Background Despite, the potential clinical utility of 60–4 visual fields, they are not frequently used in clinical practice partly, due to the purported impact of facial contour on field defects. The purpose of this study was to design and test an artificial intelligence-driven platform to predict facial structure-dependent visual field defects on 60–4 visual field tests. Methods Subjects with no ocular pathology were included. Participants were subject to optical coherence tomography, 60–4 Swedish interactive thresholding algorithm visual field tests and photography. The predicted visual field was compared with observed 60–4 visual field results in subjects. Average and point-specific sensitivity, specificity, precision, negative predictive value, accuracy, and F1-scores were primary outcome measures. Results 30 healthy were enrolled. Three-dimensional facial reconstruction using a convolution neural network (CNN) was able to predict facial contour-dependent 60–4 visual field defects in 30 subjects without ocular pathology. Overall model accuracy was 97%±3% and 96%±3% and the F1-score, dependent on precision and sensitivity, was 58%±19% and 55%±15% for the right eye and left eye, respectively. Spatial-dependent model performance was observed with increased sensitivity and precision within the far inferior nasal field reflected by an average F1-score of 76%±20% and 70%±29% for the right eye and left eye, respectively. Conclusions This pilot study reports the development of a CNN-enhanced platform capable of predicting 60–4 visual field defects in healthy controls based on facial contour. Further study with this platform may enhance understanding of the influence of facial contour on 60–4 visual field testing.