Faculty Bibliography

AIMS: To determine the retinal nerve fibre layer (RNFL) thickness at which visual field (VF) damage becomes detectable and associated with structural loss. METHODS: In a prospective cross-sectional study, 72 healthy and 40 glaucoma subjects (one eye per subject) recruited from an academic institution had VF examinations and spectral domain optical coherence tomography (SD-OCT) optic disc cube scans (Humphrey field analyser and Cirrus HD-OCT, respectively). Comparison of global mean and sectoral RNFL thicknesses with VF threshold values showed a plateau of threshold values at high RNFL thicknesses and a sharp decrease at lower RNFL thicknesses. A 'broken stick' statistical model was fitted to global and sectoral data to estimate the RNFL thickness 'tipping point' where the VF threshold values become associated with the structural measurements. The slope for the association between structure and function was computed for data above and below the tipping point. RESULTS: The mean RNFL thickness threshold for VF loss was 75.3 mum (95% CI: 68.9 to 81.8), reflecting a 17.3% RNFL thickness loss from age-matched normative value. Above the tipping point, the slope for RNFL thickness and threshold value was 0.03 dB/mum (CI: -0.02 to 0.08) and below the tipping point, it was 0.28 dB/mum (CI: 0.18 to 0.38); the difference between the slopes was statistically significant (p<0.001). A similar pattern was observed for quadrant and clock-hour analysis. CONCLUSIONS: Substantial structural loss ( approximately 17%) appears to be necessary for functional loss to be detectable using the current testing methods.

We develop an automated method to determine the foveola location in macular 3D-OCT images in either healthy or pathological conditions. Structural Support Vector Machine (S-SVM) is trained to directly predict the location of the foveola, such that the score at the ground truth position is higher than that at any other position by a margin scaling with the associated localization loss. This S-SVM formulation directly minimizes the empirical risk of localization error, and makes efficient use of all available training data. It deals with the localization problem in a more principled way compared to the conventional binary classifier learning that uses zero-one loss and random sampling of negative examples. A total of 170 scans were collected for the experiment. Our method localized 95.1% of testing scans within the anatomical area of the foveola. Our experimental results show that the proposed method can effectively identify the location of the foveola, facilitating diagnosis around this important landmark.

Optic nerve degeneration caused by glaucoma is a leading cause of blindness worldwide. Patients affected by the normal-pressure form of glaucoma are more likely to harbor risk alleles for glaucoma-related optic nerve disease. We have performed a meta-analysis of two independent genome-wide association studies for primary open angle glaucoma (POAG) followed by a normal-pressure glaucoma (NPG, defined by intraocular pressure (IOP) less than 22 mmHg) subgroup analysis. The single-nucleotide polymorphisms that showed the most significant associations were tested for association with a second form of glaucoma, exfoliation-syndrome glaucoma. The overall meta-analysis of the GLAUGEN and NEIGHBOR dataset results (3,146 cases and 3,487 controls) identified significant associations between two loci and POAG: the CDKN2BAS region on 9p21 (rs2157719 [G], OR = 0.69 [95%CI 0.63-0.75], p = 1.86x10(-)(1)(8)), and the SIX1/SIX6 region on chromosome 14q23 (rs10483727 [A], OR = 1.32 [95%CI 1.21-1.43], p = 3.87x10(-)(1)(1)). In sub-group analysis two loci were significantly associated with NPG: 9p21 containing the CDKN2BAS gene (rs2157719 [G], OR = 0.58 [95% CI 0.50-0.67], p = 1.17x10(-)(1)(2)) and a probable regulatory region on 8q22 (rs284489 [G], OR = 0.62 [95% CI 0.53-0.72], p = 8.88x10(-)(1)(0)). Both NPG loci were also nominally associated with a second type of glaucoma, exfoliation syndrome glaucoma (rs2157719 [G], OR = 0.59 [95% CI 0.41-0.87], p = 0.004 and rs284489 [G], OR = 0.76 [95% CI 0.54-1.06], p = 0.021), suggesting that these loci might contribute more generally to optic nerve degeneration in glaucoma. Because both loci influence transforming growth factor beta (TGF-beta) signaling, we performed a genomic pathway analysis that showed an association between the TGF-beta pathway and NPG (permuted p = 0.009). These results suggest that neuro-protective therapies targeting TGF-beta signaling could be effective for multiple forms of glaucoma.

PURPOSE: To develop an automated method to identify the normal macula and three macular pathologies (macular hole [MH], macular edema [ME], and age-related macular degeneration [AMD]) from the fovea-centered cross sections in three-dimensional (3D) spectral-domain optical coherence tomography (SD-OCT) images. METHODS: A sample of SD-OCT macular scans (macular cube 200 x 200 or 512 x 128 scan protocol; Cirrus HD-OCT; Carl Zeiss Meditec, Inc., Dublin, CA) was obtained from healthy subjects and subjects with MH, ME, and/or AMD (dataset for development: 326 scans from 136 subjects [193 eyes], and dataset for testing: 131 scans from 37 subjects [58 eyes]). A fovea-centered cross-sectional slice for each of the SD-OCT images was encoded using spatially distributed multiscale texture and shape features. Three ophthalmologists labeled each fovea-centered slice independently, and the majority opinion for each pathology was used as the ground truth. Machine learning algorithms were used to identify the discriminative features automatically. Two-class support vector machine classifiers were trained to identify the presence of normal macula and each of the three pathologies separately. The area under the receiver operating characteristic curve (AUC) was calculated to assess the performance. RESULTS: The cross-validation AUC result on the development dataset was 0.976, 0.931, 0939, and 0.938, and the AUC result on the holdout testing set was 0.978, 0.969, 0.941, and 0.975, for identifying normal macula, MH, ME, and AMD, respectively. CONCLUSIONS: The proposed automated data-driven method successfully identified various macular pathologies (all AUC > 0.94). This method may effectively identify the discriminative features without relying on a potentially error-prone segmentation module.

We address a novel problem domain in the analysis of optical coherence tomography (OCT) images: the diagnosis of multiple macular pathologies in retinal OCT images. The goal is to identify the presence of normal macula and each of three types of macular pathologies, namely, macular edema, macular hole, and age-related macular degeneration, in the OCT slice centered at the fovea. We use a machine learning approach based on global image descriptors formed from a multi-scale spatial pyramid. Our local features are dimension-reduced local binary pattern histograms, which are capable of encoding texture and shape information in retinal OCT images and their edge maps, respectively. Our representation operates at multiple spatial scales and granularities, leading to robust performance. We use 2-class support vector machine classifiers to identify the presence of normal macula and each of the three pathologies. To further discriminate sub-types within a pathology, we also build a classifier to differentiate full-thickness holes from pseudo-holes within the macular hole category. We conduct extensive experiments on a large dataset of 326 OCT scans from 136 subjects. The results show that the proposed method is very effective (all AUC>0.93).

Aqueous humor (AH) exiting the eye via the trabecular meshwork and Schlemm's canal (SC) passes through the deep and intrascleral venous plexus (ISVP) or directly through aqueous veins. The purpose of this study was to visualize the human AH outflow system 360 degrees in three dimensions (3D) during active AH outflow in a virtual casting. The conventional AH outflow pathways of 7 donor eyes were imaged with a modified Bioptigen spectral-domain optical coherence tomography system (Bioptigen Inc, USA; SuperLum LTD, Ireland) at a perfusion pressure of 20 mmHg (N = 3), and 10 mmHg (N = 4). In all eyes, 36 scans (3 equally distributed in each clock hour), each covering a 2 x 3 x 2 mm volume (512 frames, each 512 x 1024 pixels), were obtained. All image data were black/white inverted, and the background subtracted (ImageJ 1.40 g, http://rsb.info.nih.gov/ij/). Contrast was adjusted to isolate the ISVP. SC, collector channels, the deep and ISVP, and episcleral veins were observed throughout the limbus. Aqueous veins could be observed extending into the episcleral veins. Individual scan ISVP castings were rendered and assembled in 3D space in Amira 4.1 (Visage Imaging Inc. USA). A 360-degree casting of the ISVP was obtained in all perfused eyes. The ISVP tended to be dense and overlapping in the superior and inferior quadrants, and thinner in the lateral quadrants. The human AH outflow pathway can be imaged using SD-OCT. The more superficial structures of the AH outflow pathway present with sufficient contrast as to be optically isolated and cast in-situ 360 degrees in cadaver eye perfusion models. This approach may be useful as a model in future studies of human AH outflow.