Faculty Bibliography

Purpose: Macular ganglion cell-inner plexiform layer (GCIPL) measurement with OCT has been suggested as an alternative for detecting progression in advanced stages of glaucoma. In this study we evaluated if conventional optic nerve head (ONH) and visual field (VF) parameters could be used to detect progression in eyes with advanced structural damage, in which the GCIPL approaches the practical minimal measurable level (floor effect).
Method(s): Subjects with advanced structural glaucoma (average circumpapillary retinal nerve fiber layer (cRNFL) thickness <=60mum) with >= 4 visits with qualified perimetry (Humphrey Field Analyzer; Zeiss) and spectral-domain OCT (Cirrus HD-OCT; Zeiss) were enrolled. Subjects were divided into Change or No change groups based on their GCIPL findings on macular guided progression analysis (GPA). No change was defined as: no change >2mum in average, superior or inferior GCIPL between the first and last visits, no statistically significant rate of change, and no cluster marking change in deviation maps in any visits. Structural (cRNFL, GCIPL and ONH) and functional (VF mean deviation (MD) and visual field index (VFI)) parameters were analyzed using a hierarchical linear model adjusting for eye correlation within subjects. Covariates included were age, ethnicity, signal strength, inclusion in No change group, follow-up duration and its interaction with the No change group.
Result(s): 44 eyes (37 subjects) qualified for the study with 22 eyes (50%) in each group. The average age at baseline was 67.0+/-11.4 years and mean follow-up was 4.1+/-1.8 years. The Change group had significantly thicker average, superior and inferior GCIPLs compared to the No change group at baseline (Table 1). Longitudinal analysis showed significant rates of change for most parameters in the Change group (Table 2). In the No change group, no significant thinning of the cRNFL was detected whereas VF and ONH parameters (Cup/Disc ratios and cup volume) showed significant change.
Conclusion(s): Eyes with advanced glaucomatous structural damage approaching the presumable GCIPL floor effect on OCT demonstrated changes in ONH and VF parameters even when no further RNFL thinning was detected. Novel parameters for evaluating ONH structure may be useful in the follow-up of advanced glaucoma. (Table presented)

Purpose: The optic nerve and peripapillary sclera undergo mechanical stresses and strains due to tractional forces as the eyes move. In this study, gaze as a potential cause of lamina cribrosa (LC) deformation was explored in a well-controlled in-vivo animal model at normal and elevated intracranial pressure (ICP).
Method(s): An adult healthy macaque was anesthetized, and OCT (Leica Microsystems, Chicago, IL) scans of the optic nerve head (ONH) (3x3mm; 400x400x1024 pixels) were obtained. A baseline scan was acquired at normal ICP (9mmHg) with the eye at neutral position followed by adduction and abduction positions. ICP was raised to 25mmHg via a ventricular cannula, and scanning was repeated in all gaze settings and locations. All scans were acquired after a 10-minute pause to allow for dissipation of tissue viscoelastic changes. Scans were registered in 3D using our own algorithm and evaluated for macroand microstructure deformation. Lamina microstructure measurements were generated from shared regions among all scanning setting using our own 3D segmentation algorithm.
Result(s): At baseline and elevated ICPs, the IOPs were10 and 19mmHg, respectively. Gaze shifts from the neutral position were associated with a seesaw movement of the macrostructure - nasal elevation and temporal depression in adduction and the reverse effect in abduction (Fig. 1). This effect was more pronounced in elevated ICP condition. At both pressure settings, the ratio of beam thickness to pore diameter increased when gaze deviated from midline (Table 1). The changes seen from neutral to abduction were greater than those seen from neutral to adduction; both findings were more pronounced under elevated ICP.
Conclusion(s): We demonstrated that gaze can induce noticeable macrostructure deformation of the ONH region and a measurable effect on global LC microstructural parameters. Microstructure effects are more pronounced in abduction and in elevated ICP. The magnitude of gaze effect as well as the potential damage to the lamina and its associated axons should be studied further. (Figure presented)

Purpose: It has been shown that Asians have a high prevalence of normal tension glaucoma, while Caucasians and African-Americans have predominantly high tension glaucoma. This study compared the lamina cribrosa (LC) microstructure of a cohort of Korean and American eyes in order to discern microstructure differences between the cohorts that could contribute to this phenomenon.
Method(s): The optic nerve head of 53 healthy eyes (42 subjects; 38 Korean eyes and 15 American eyes, consisting of a mixture of Caucasian and African-American eyes) was scanned 3 times during the same session with Cirrus HD-OCT (Zeiss, Dublin CA). These scans were registered and averaged to increase LC visibility using a method we have previously described. The area of the ONH featuring clearly visible LC was delineated, and scans were semi-automatically analyzed within the delineated area to segment the LC microstructure in 3D. The LC measurements of beam thickness, pore diameter, and beam/pore ratio were compared using a hierarchical linear model taking ethnicity and age into account.
Result(s): Baseline characteristics were similar between the cohorts (Table 1). Mean pore diameter was on average 3.78mum bigger in Asian subjects compared to non-Asian subjects (p<0.001), and beam/pore ratio was 0.33 units smaller in Asian subjects (p<0.001). No differences were detected in beam thickness.
Conclusion(s): The in vivo microstructure of the LC varies among different ethnicities. Further research is needed to determine the cause, effect and clinical relevance of these differences. (Table presented)

Purpose : We previously described the two-dimensional continuous time hidden Markov model (2D CT-HMM) to model glaucoma progression using structural and functional measurements simultaneously. The purpose of this study was to validate the glaucoma progression prediction performance of a previously trained model on data collected from a different cohort. Methods : A 2D CT-HMM was trained using optical coherence tomography (OCT; Cirrus HD-OCT, Zeiss, Dublin, CA) mean circumpapillary retinal nerve fiber layer (cRNFL) thickness and visual field index (VFI; Humphrey Field Analyzer, Zeiss) obtained from 107 eyes of 107 subjects, including glaucoma and glaucoma suspect. Average observation period was 4.2 years (7.1 visits). Approximately 1 year of longitudinal data were collected from a separate cohort. 78 eyes of 39 subjects, glaucoma and glaucoma suspect, with an average of 2.2 +/- 0.4 visits were included. After matching the distribution of OCT and VF data on the training cohort, 19 eyes from 14 subjects were selected. The previously trained model was tested on these cases. One visit was used as an input to the model to predict the state at the next visit at least 6 months later, with 4 possible state changes (stable, OCT, VF, or OCT+VF progression). The percentage of correct prediction against the actual recorded state was reported as the prediction accuracy. Results : Baseline age of the test cohort was 58.4 +/- 13.9 years, VFI 93.6 +/- 8.3, mean cRNFL thickness 74.0 +/- 10.9mum. Figure 1 shows the trained model. The size of the circle (state) shows the number of subjects passing through the state. The grayscale of the state indicates the length of time spent there, increasing white to black. Lines indicate state changes, with the blue line being the most likely. This information is also shown in numerical form. The inset shows an example of model use. The calculated prediction accuracy of this pre-trained 2D CT-HMM on test data was 52.6%. Conclusions : Although the glaucoma progression prediction performance of the trained 2D CT-HMM was slightly lower than that previously reported, it is acceptable given the training and testing cohorts were different, and it exceeds the random chance of making a correct prediction, 25%. Furthermore, unlike conventional methods, this model requires only one visit as an input, which makes it a potentially useful tool in the clinical prediction of glaucoma progression. (Figure Presented)

Purpose: When assessing treatment effects in observational studies, the propensity score (PS) method is commonly used to reduce the selection bias of treatments. Weighting subjects by the inverse probability of treatment using the PS mimics treatment s ran e domization e Our y c o ntin as to g apply PS t r website, you are ag ation o g glaucoma treatment c cmeepdtication with rates of structural changes in a longitudinal cohort of glaucoma subjects.
Method(s): Glaucoma subjects treated with prostaglandin, beta blockers, and/or carbonic anhydrase inhibitors (CAIs) with > 2 visits with qualified OCT (Cirrus HD-OCT; Zeiss) were included. Subjects were on medication for at least 3 months prior to each OCT visit. Multinomial PS for baseline medication selection was estimated by baseline age, visual field (VF) mean deviation (MD), intraocular pressure and ethnicity. Rates of change for OCT's average circumpapillary retinal nerve fiber layer (RNFL) and macular ganglion cell-inner plexiform layer (GCIPL) thicknesses were calculated per eye using linear regression. Their associations with baseline RNFL, GCIPL, baseline medication, post-baseline medication and medication duration were tested using linear regression with and without PS weighting.
Result(s): 207 eyes (117 subjects) were qualified with average age of 62.2+/-12.7 years and median MD of -3.6 dB (IQR -9.0, -1.4) at baseline, and a mean follow-up of 3.2+/-1.8 years. The average duration of treatment range from 1.3+/-1.8 to 2.4+/-2.5 years for CAIs and prostaglandin, respectively. At baseline, average RNFL and GCIPL were 71.5+/-14.4 mum and 65.9+/-13.5 mum. Throughout follow-up, mean rate of change for RNFL and GCIPL were -0.30+/-2.60 mum/year and 0.27+/-7.72 mum/year. Without PS weighting, no medication effect was shown to be associated with either rate of change. With PS weighting, however, the rate of change for RNFL was significantly associated with taking CAIs (-1.26 mum/year, p=0.029) and prostaglandin (-0.98 mum/year, p=0.044) and baseline RNFL (-0.05 mum/year, p=0.017). Longer use of the medications slowed RNFL decrease, although the effects were not statistically significant. No association was detected between treatment and rate of change for GCIPL.
Conclusion(s): PS can be useful to reduce treatment selection bias and facilitate more rigorous estimation of medication effects in observational glaucoma research

The lamina cribrosa is a primary site of damage in glaucoma. While mechanical distortion is hypothesized to cause reduction of axoplasmic flow, little is known about how the pores, which contains the retinal ganglion cell axons, traverse the lamina cribrosa. We investigated lamina cribrosa pore paths in vivo to quantify differences in tortuosity of pore paths between healthy and glaucomatous eyes. We imaged 16 healthy, 23 glaucoma suspect and 48 glaucomatous eyes from 70 subjects using a swept source optical coherence tomography system. The lamina cribrosa pores were automatically segmented using a previously described segmentation algorithm. Individual pore paths were automatically tracked through the depth of the lamina cribrosa using custom software. Pore path convergence to the optic nerve center and tortuosity was quantified for each eye. We found that lamina cribrosa pore pathways traverse the lamina cribrosa closer to the optic nerve center along the depth of the lamina cribrosa regardless of disease severity or diagnostic category. In addition, pores of glaucoma eyes take a more tortuous path through the lamina cribrosa compared to those of healthy eyes, suggesting a potential mechanism for reduction of axoplasmic flow in glaucoma.

The choroid is vascular tissue located underneath the retina and supplies oxygen to the outer retina; any damage to this tissue can be a precursor to retinal diseases. This paper presents an automated method of choroidal segmentation from Enhanced Depth Imaging Optical Coherence Tomography (EDI-OCT) images. The Dijkstra shortest path algorithm is used to segment the choroid-sclera interface (CSI), the outermost border of the choroid. A novel intensity-normalisation technique that is based on the depth of the choroid is used to equalise the intensity of all non-vessel pixels in the choroid region. The outer boundary of choroidal vessel and CSI are determined approximately and incorporated to the edge weight of the CSI segmentation to choose optimal edge weights. This method is tested on 190 B-scans of 10 subjects against choroid thickness (CTh) results produced manually by two graders. For comparison, results obtained by two state-of-the-art automated methods and our proposed method are compared against the manual grading, and our proposed method performed the best. The mean root-mean-square error (RMSE) for finding the CSI boundary by our method is 7.71±6.29 pixels, which is significantly lower than the RMSE for the two other state-of-the-art methods (36.17±11.97 pixels and 44.19±19.51 pixels). The correlation coefficient for our method is 0.76, and 0.51 and 0.66 for the other two state-of-the-art methods. The interclass correlation coefficients are 0.72, 0.43 and 0.56 respectively. Our method is highly accurate, robust, reliable and consistent. This identification can enable to quantify the biomarkers of the choroidin large scale study for assessing, monitoring disease progression as well as early detection of retinal diseases. Identification of the boundary can help to determine the loss or change of choroid, which can be used as features for the automatic determination of the stages of retinal diseases.

In this paper, we propose a novel classification model for automatically identifying individuals with age-related macular degeneration (AMD) or Diabetic Macular Edema (DME) using retinal features from Spectral Domain Optical Coherence Tomography (SD-OCT) images. Our classification method uses retinal features such as the thickness of the retina and the thickness of the individual retinal layers, and the volume of the pathologies such as drusen and hyper-reflective intra-retinal spots. We extract automatically, ten clinically important retinal features by segmenting individual SD-OCT images for classification purposes. The effectiveness of the extracted features is evaluated using several classification methods such as Random Forrest on 251 (59 normal, 177 AMD and 15 DME) subjects. We have performed 15-fold cross-validation tests for three phenotypes; DME, AMD and normal cases using these data sets and achieved accuracy of more than 95% on each data set with the classification method using Random Forrest. When we trained the system as a two-class problem of normal and eye with pathology, using the Random Forrest classifier, we obtained an accuracy of more than 96%. The area under the receiver operating characteristic curve (AUC) finds a value of 0.99 for each dataset. We have also shown the performance of four state-of-the-methods for classification the eye participants and found that our proposed method showed the best accuracy.

Glaucoma is a leading cause of blindness that leads to characteristic changes in the optic nerve head (ONH) region, such as nasalization of vessels. It is unknown whether the spatial location of this vessel shift inside the ONH occurs within the lamina cribrosa (LC) or the prelaminar tissue. The purpose of this study was to compare the location of the central retinal vessel trunk (CRVT) in the LC and prelaminar tissue in living healthy and glaucomatous eyes. We acquired 3-dimensional ONH scans from 119 eyes (40 healthy, 29 glaucoma suspect, and 50 glaucoma) using optical coherence tomography (OCT). The CRVT location was manually delineated in separate projection images of the LC and prelamina. We found that the CRVT in glaucoma suspect and glaucomatous eyes was located significantly more nasally compared to healthy eyes at the level of the prelamina. There was no detectable difference found in the location of the CRVT at the level of the LC between diagnostic groups. While the nasal location of the CRVT in the prelamina has been associated with glaucomatous axonal death, our results suggest that the CRVT in the LC is anchored in the tissue with minimal variation in glaucomatous eyes.

Purpose: Guided progression analysis (GPA) is a commonly used glaucoma progression detection method based on optical coherence tomography (OCT) that measures retinal nerve fiber layer (RNFL) thickness obtained with Cirrus HD-OCT. Recently, GPA based on ganglion cell inner plexiform layer (GCIPL) was introduced. The purpose of this study was to assess the agreement in progression detection between GPA using RNFL and GCIPL measurements. Methods: 118 open angle glaucoma eyes (78 subjects), 50 glaucoma suspects eyes (28 subjects), and 4 healthy eyes (2 subjects) that had comprehensive ophthalmic examination and greater than or equal to 5 visits with qualified OCT scans of the macula and optic nerve head regions were enrolled. GPA was used in all eyes with matching dates for baseline and final visits for RNFL and GCIPL analysis. Considering that trend analyses for both regions was previously reported, we focused on the event analysis where "probable event" of progression was defined as the first test showing progression and "likely event" as the one that immediately followed the probable event also showing progression. Stuart-Maxwell test was used for assessing agreement in the categorical analysis of progression. Results: Mean subject age was 68.5 +/- 10.2 years and median baseline visual field mean deviation was -1.5dB ([Q1, Q3]; -4.32, -0.13). The majority of the eyes did not progress, but progression agreement between average RNFL and GCIPL and for superior RNFL and GCIPL showed statistically significant differences (P=0.017 and P<0.001, respectively; Table 1). No difference was detected in agreement for progression between inferior RNFL and GCIPL (P=0.389). Conclusions: Superior and inferior RNFL and GCIPL GPAs showed limited agreement in detecting progression. Further investigation is required to identify the factors affecting this disparity