ResearchPad - glaucoma https://www.researchpad.co Default RSS Feed en-us © 2020 Newgen KnowledgeWorks <![CDATA[Mutations in <i>SPATA13/ASEF2</i> cause primary angle closure glaucoma]]> https://www.researchpad.co/article/elastic_article_15764 Glaucoma is the leading cause of irreversible blindness globally. Angle closure glaucoma accounts for 50% of all glaucoma blindness impacting quality of life and burden on health services. A number of variations in DNA appear to influence the risk of the disease. However, the biological mechanism underlying this important disease remains unclear. In this paper, we report the identification and functional characterisation of the first gene, mutation in which causes primary angle closure glaucoma in a seven generation Caucasian family. We have identified other variants in the same gene in another family and individuals with the disease. This gene is involved in cell division and is highly expressed in parts of the eye affected by the disease. Mutations in this gene appear to affect important enzyme activity involved in cell division. Identification of the disease-causing role of mutations in this gene helps to further the understanding of glaucoma aetiology and identifies potential therapeutic targets for disease management.

]]>
<![CDATA[Hemisphere opposite to vascular trunk deviation is earlier affected by glaucomatous damage in myopic high-tension glaucoma]]> https://www.researchpad.co/article/elastic_article_15760 To investigate whether the position of the central vascular trunk, as a surrogate of lamina cribrosa (LC) shift, is associated with the initial hemisphere of visual field defect in myopic high-tension glaucoma (HTG) eyes.MethodsThe deviation of the central vascular trunk was measured from the center of the Bruch’s membrane opening (BMO), which was delineated by OCT imaging. The angular deviation was measured with the horizontal nasal midline as 0° and the superior location as a positive value. The initial hemisphere developing visual field defect was defined as three connected abnormal points (having a P value with less than 0.5% probability of being normal) appearing in only one hemisphere in pattern deviation plots. If those points were observed in both hemispheres initially, the eye was classified as bi-hemispheric visual field defect.ResultsInitially, 36 eyes (44%) had superior visual field defects, 27 (33%) inferior visual field defects, and 18 (22%) bi-hemispheric visual field defects. After a mean follow-up of 5 years, the number of bi-hemispheric visual field defects had increased to 34 (42%). A logistic regression analysis revealed that inferior deviation of vascular trunk was the only factor associated with initial inferior visual field defect (P = 0.001), while initial bi-hemispheric visual field defects were associated with worse mean deviation at initial visits (P<0.001). A conditional inference tree analysis showed that both the angular deviation (P<0.001) and initial mean deviation (P = 0.025) determined the initial hemispheres developing visual field defect.ConclusionsAlthough both hemispheres were involved as glaucoma progression, the axons on the side counter to the vascular trunk deviation were damaged earlier in HTG. This finding implies the LC shift could add additional stress to axons exposed to high intraocular pressure. ]]> <![CDATA[3D-Reconstruction of the human conventional outflow system by ribbon scanning confocal microscopy]]> https://www.researchpad.co/article/elastic_article_15723 The risk for glaucoma is driven by the microanatomy and function of the anterior segment. We performed a computation-intense, high-resolution, full-thickness ribbon-scanning confocal microscopy (RSCM) of the outflow tract of two human eyes. We hypothesized this would reveal important species differences when compared to existing data of porcine eyes, an animal that does not spontaneously develop glaucoma.MethodsAfter perfusing two human octogenarian eyes with lectin-fluorophore conjugate and optical clearance with benzyl alcohol benzyl benzoate (BABB), anterior segments were scanned by RSCM and reconstructed in 3D for whole-specimen rendering. Morphometric analyses of the outflow tract were performed for the trabecular meshwork (TM), limbal, and perilimbal outflow structures and compared to existing porcine data.ResultsRSCM provided high-resolution data for IMARIS-based surface reconstruction of outflow tract structures in 3D. Different from porcine eyes with an abundance of highly interconnected, narrow, and short collector channels (CCs), human eyes demonstrated fewer CCs which had a 1.5x greater cross-sectional area (CSA) and 2.6x greater length. Proximal CC openings at the level of Schlemm’s canal (SC) had a 1.3x larger CSA than distal openings into the scleral vascular plexus (SVP). CCs were 10.2x smaller in volume than the receiving SVP vessels. Axenfeld loops, projections of the long ciliary nerve, were also visualized.ConclusionIn this high-resolution, volumetric RSCM analysis, human eyes had far fewer outflow tract vessels than porcine eyes. Human CCs spanned several clock-hours and were larger than in porcine eyes. These species differences may point to factors downstream of the TM that increase our vulnerability to glaucoma. ]]> <![CDATA[Deep learning assisted detection of glaucomatous optic neuropathy and potential designs for a generalizable model]]> https://www.researchpad.co/article/elastic_article_14620 To evaluate ways to improve the generalizability of a deep learning algorithm for identifying glaucomatous optic neuropathy (GON) using a limited number of fundus photographs, as well as the key features being used for classification.MethodsA total of 944 fundus images from Taipei Veterans General Hospital (TVGH) were retrospectively collected. Clinical and demographic characteristics, including structural and functional measurements of the images with GON, were recorded. Transfer learning based on VGGNet was used to construct a convolutional neural network (CNN) to identify GON. To avoid missing cases with advanced GON, an ensemble model was adopted in which a support vector machine classifier would make final classification based on cup-to-disc ratio if the CNN classifier had low-confidence score. The CNN classifier was first established using TVGH dataset, and then fine-tuned by combining the training images of TVGH and Drishti-GS datasets. Class activation map (CAM) was used to identify key features used for CNN classification. Performance of each classifier was determined through area under receiver operating characteristic curve (AUC) and compared with the ensemble model by diagnostic accuracy.ResultsIn 187 TVGH test images, the accuracy, sensitivity, and specificity of the CNN classifier were 95.0%, 95.7%, and 94.2%, respectively, and the AUC was 0.992 compared to the 92.8% accuracy rate of the ensemble model. For the Drishti-GS test images, the accuracy of the CNN, the fine-tuned CNN and ensemble model was 33.3%, 80.3%, and 80.3%, respectively. The CNN classifier did not misclassify images with moderate to severe diseases. Class-discriminative regions revealed by CAM co-localized with known characteristics of GON.ConclusionsThe ensemble model or a fine-tuned CNN classifier may be potential designs to build a generalizable deep learning model for glaucoma detection when large image databases are not available. ]]> <![CDATA[Diagnostic power of scleral spur length in primary open-angle glaucoma]]> https://www.researchpad.co/article/elastic_article_9777 To investigate the diagnostic capability of scleral spur length in discriminating eyes with primary open-angle glaucoma (POAG) from healthy eyes.MethodsSeventy-eight eyes of 78 patients with POAG and 93 eyes of 93 age-, sex- and axial length-matched healthy subjects were included. The scleral spur length was measured using swept-source optical coherence tomography. Receiver operating characteristic (ROC) curves were derived based on the measurements.ResultsThe scleral spur length was significantly shorter in POAG eyes compared with healthy eyes (Method I, 164.91 ± 23.36 vs. 197.60 ± 25.32 μm; Method II, 145.15 ± 16.59 vs. 166.95 ± 19.31 μm; Method III, 162.33 ± 22.83 vs. 185.12 ± 23.58 μm, respectively; all p < 0.001). The areas under ROC curves were 0.841 (Method I), 0.810 (Method II), and 0.753 (Method III) for the scleral spur length. Moreover, Schlemm’s canal area was significantly associated with the scleral spur length (Method I) in both POAG (β = 0.027; p < 0.001) and healthy (β = 0.016; p = 0.009) groups.ConclusionsThe scleral spur length had a good discriminating capability between POAG and healthy eyes, and it could be a novel biomarker for POAG evaluation clinically. ]]> <![CDATA[Neuroprotective effects of exogenous erythropoietin in Wistar rats by downregulating apoptotic factors to attenuate N-methyl-D-aspartate-mediated retinal ganglion cells death]]> https://www.researchpad.co/article/N85685bba-c047-422b-abfc-358a98ed1fe7

The aim of this study was to investigate whether exogenous erythropoietin (EPO) administration attenuates N-methyl-D-aspartate (NMDA)-mediated excitotoxic retinal damage in Wistar rats. The survival rate of retinal ganglion cells (RGCs) were investigated by flat mount analysis and flow cytometry. A total of 125 male Wistar rats were randomly assigned to five groups: negative control, NMDA80 (i.e., 80 nmoles NMDA intravitreally injected), NMDA80 + 10ng EPO, NMDA80 + 50ng EPO, and NMDA80 + 250ng EPO. The NMDA80 + 50ng EPO treatment group was used to evaluate various administrated points (pre-/co-/post- administration of NMDA80). Meanwhile, the transferase dUTP Nick-End Labeling (TUNEL) assay of RGCs, the inner plexiform layer (IPL) thickness and the apoptotic signal transduction pathways of μ-calpain, Bax, and caspase 9 were assessed simultaneously using an immunohistochemical method (IHC). When EPO was co-administered with NMDA80, attenuated cell death occurred through the downregulation of the apoptotic indicators: μ-calpain was activated first (peak at ~18hrs), followed by Bax and caspase 9 (peak at ~40hrs). Furthermore, the images of retinal cross sections have clearly demonstrated that thickness of the inner plexiform layer (IPL) was significantly recovered at 40 hours after receiving intravitreal injection with NMDA80 and 50ng EPO. Exogenous EPO may protect RGCs and bipolar cell axon terminals in IPL by downregulating apoptotic factors to attenuate NMDA-mediated excitotoxic retinal damage.

]]>
<![CDATA[Structure-Function Agreement Is Better Than Commonly Thought in Eyes With Early Glaucoma]]> https://www.researchpad.co/article/Ncaecd3b3-def9-469a-8ec5-dcb7d6f67293

Purpose

To assess the agreement between structural (optical coherence tomography [OCT]) and functional (visual field [VF]) glaucomatous damage with an automated method and deviation/probability maps, and to compare this method to a metric method.

Methods

Wide-field spectral-domain OCT scans, including the disc and macula, and 24-2 and 10-2 VFs were obtained from 45 healthy control (H) eyes/individuals, and 53 eyes/patients with 24-2 mean deviation (MD) better than −6 dB diagnosed as “definite glaucoma” (DG) by experts. Abnormal structure–abnormal function (aS-aF) agreement was assessed with an automated topographic (T) method based upon VF pattern deviation and OCT probability maps. Results were compared to a metric (M) method optimized for accuracy, (abnormal 24-2 glaucoma hemifield test [GHT] or pattern standard deviation [PSD], or 10-2 PSD AND abnormal OCT [quadrant]).

Results

For the T-method, 47 (88.7%) of the 53 DG eyes showed aS-aF agreement, compared to 2 (4.5%) of the 45 H eyes. The aS-aF agreement for these two H eyes was easily identified as mistaken, and did not replicate on a subsequent test. Without the 10-2, the aS-aF agreement decreased from 47 to 34 (64.2%) of 53 DG eyes. For the M-method, 37 (69.8%) of the 53 DG eyes showed aS-aF agreement, while omitting the 10-2 VF resulted in agreement in only 33 (62.3%) eyes.

Conclusions

There is good agreement between structural and functional damage, even in eyes with confirmed early glaucomatous damage, if both 24-2 and 10-2 VFs are obtained, and abnormal locations on the VFs are compared to abnormal regions seen on OCT macular and disc scans. This can be done in an objective, automated fashion. (ClinicalTrials.gov number, NCT02547740.)

]]>
<![CDATA[Latanoprost could exacerbate the progression of presbyopia]]> https://www.researchpad.co/article/5c5ca29cd5eed0c48441e76d

Purpose

Prostaglandin analogues (PG) reduce intra-ocular pressure by enhancing uveoscleral flow at the ciliary body, which controls accommodation via the ciliary muscle. We investigated the effect of PG on accommodation and presbyopia progression in glaucoma patients.

Methods

We conducted a clinic-based, retrospective, cross-sectional study. Inclusion criteria were bilateral phakic patients aged 40–69 years with best corrected visual acuity better than 20/30. Exclusion criteria were any disease affecting vision other than glaucoma and history of ocular surgery. Subjects with no prescription or vision-affecting disease served as controls (n = 260). The glaucoma patients were prescribed eye drops containing 0.005% latanoprost for more than six months (n = 23). We measured the binocular near add power at a distance of 30 cm in both groups and compared the results using Kaplan-Meier analysis.

Results

The mean age (± SD) of the control subjects was 51.5 ± 5.2 years and 39% were male. Similarly, the glaucoma patients had a mean age of 51.0 ± 7.2 years and 39% were male. There were no significant differences in age, gender, intra-ocular pressure, spherical equivalent, astigmatism, or anisometropia between groups. Survival analysis indicated that the glaucoma patients in this study reached the endpoint (near add power of +3.00 D) significantly earlier than control patients (P = 0.0001; generalized Wilcoxon test).

Conclusions

Exacerbation of presbyopia progression in glaucoma patients is a potential side effect of latanoprost eyedrops.

]]>
<![CDATA[Neuroprotective and neuroregenerative effects of CRMP-5 on retinal ganglion cells in an experimental in vivo and in vitro model of glaucoma]]> https://www.researchpad.co/article/5c521853d5eed0c484797c19

Purpose

To analyze the potential neuro-protective and neuro-regenerative effects of Collapsin-response-mediator-protein-5 (CRMP-5) on retinal ganglion cells (RGCs) using in vitro and in vivo animal models of glaucoma.

Methods

Elevated intraocular pressure (IOP) was induced in adult female Sprague-Dawley (SD) rats by cauterization of three episcleral veins. Changes in CRMP-5 expression within the retinal proteome were analyzed via label-free mass spectrometry. In vitro, retinal explants were cultured under elevated pressure (60 mmHg) within a high-pressure incubation chamber with and without addition of different concentrations of CRMP-5 (4 μg/l, 200 μg/l and 400 μg/l). In addition, retinal explants were cultured under regenerative conditions with and without application of 200 μg/l CRMP-5 after performing an optic nerve crush (ONC). Thirdly, an antibody against Protein Kinase B (PKB) was added to examine the possible effects of CRMP-5. RGC count was performed. Number and length of the axons were determined and compared. To undermine a signal-transduction pathway via CRMP-5 and PKB microarray and immunohistochemistry were performed.

Results

CRMP-5 was downregulated threefold in animals showing chronically elevated IOP. The addition of CRMP-5 to retinal culture significantly increased RGC numbers under pressure in a dose-dependent manner and increased and elongated outgrowing axons in retinal explants significantly which could be blocked by PKB. Especially the number of neurites longer than 400 μm significantly increased after application of CRMP-5. CRMP-5 as well as PKB were detected higher in the experimental than in the control group.

Conclusion

CRMP-5 seems to play an important role in an animal model of glaucoma. Addition of CRMP-5 exerts neuro-protective and neuro-regenerative effects in vitro. This effect could be mediated via activation of PKB affecting intra-cellular apoptosis pathways.

]]>
<![CDATA[Mouse model of ocular hypertension with retinal ganglion cell degeneration]]> https://www.researchpad.co/article/5c466550d5eed0c4845188e3

Objectives

Ocular hypertension is a primary risk factor for glaucoma and results in retinal ganglion cell (RGC) degeneration. Current animal models of glaucoma lack severe RGC cell death as seen in glaucoma, making assessment of physiological mediators of cell death difficult. We developed a modified mouse model of ocular hypertension whereby long-lasting elevation of intraocular pressure (IOP) is achieved, resulting in significant reproducible damage to RGCs.

Results

In this model, microbeads are mixed with hyaluronic acid and injected into the anterior chamber of C57BL/6J mice. The hyaluronic acid allows for a gradual release of microbeads, resulting in sustained blockage of Schlemm’s canal. IOP elevation was bimodal during the course of the model’s progression. The first peak occurred 1 hours after beads injection, with an IOP value of 44.69 ± 6.00 mmHg, and the second peak occurred 6–12 days post-induction, with an IOP value of 34.91 ± 5.21 mmHg. RGC damage was most severe in the peripheral retina, with a loss of 64.1% compared to that of untreated eyes, while the midperiphery exhibited a 32.4% loss, 4 weeks following disease induction.

Conclusions

These results suggest that sustained IOP elevation causes more RGC damage in the periphery than in the midperiphery of the retina. This model yields significant and reproducible RGC degeneration.

]]>
<![CDATA[Long-term probability of intraocular pressure elevation with the intravitreal dexamethasone implant in the real-world]]> https://www.researchpad.co/article/5c390b97d5eed0c48491d656

Purpose

To evaluate the long-term cumulative probability of intraocular pressure (IOP) elevation with the intravitreal dexamethasone implant (IDI) when used to treat different indications: diabetic macular edema, uveitis, retinal vein occlusion.

Methods

705 IDI injections (429 eyes) were assessed and Kaplan-Meier graphs were generated to assess: the probability of different levels of IOP elevation (IOP≥21, ≥25 or ≥35 mmHg), IOP change ≥10 mmHg, initiation of IOP-lowering treatment, glaucoma surgery, IOP change with repeat injections and IOP elevation in eyes with glaucoma and ocular hypertension (OHT).

Results

The cumulative probability of IOP ≥21, ≥25 and ≥35 mmHg was 50%-60%, 25%-30% and 6%-7% at 12–24 months, respectively. The probability of initiating IOP-lowering medication was 31%-54% at 12–24 months. Glaucoma and OHT eyes had a higher probability of mild IOP elevation (≥21 mmHg, 65.1%, 75% and 57.8%, p = 0.01), yet a similar moderate (≥25 mmHg, 22.3%, 28% and 30.2%, p = 0.91) and severe elevation of IOP (≥35 mmHg, 3.7%, 7.1% and 4%, p = 0.71) as normal eyes. Glaucoma surgery was required in only 0.9% cases (4/429). At baseline, 8.8% of the treated eyes had glaucoma, 6.7% OHT and 16.9% were already on IOP-lowering medication.

Conclusions

In the long-term (24 months), IOP elevation is common, generally mild (30% IOP, ≥25 mmHg) and well-tolerated, resolving with topical treatment (54%) and rarely requiring surgery (0.9%).

]]>
<![CDATA[Post-phacoemulsification iris changes in eyes with glaucoma or glaucoma suspect status]]> https://www.researchpad.co/article/5c1c0b0dd5eed0c48442732d

Purpose

This prospective study used anterior segment optical coherence tomography (AS-OCT) to determine how phacoemulsification (phaco) changes iris parameters in eyes with glaucoma or glaucoma suspect status.

Methods

Using Visante AS-OCT (Carl Zeiss Meditec AG), the following pre- and post-phaco parameters were measured: IT750 = iris thickness at 750 μm from the scleral spur; IT2000 = iris thickness 2000 μm from the scleral spur; ITCM = the maximum iris thickness at the middle one third of the iris; ICURV = iris curvature; IAREA = iris area; and pupil size = pupil diameter (mm). Only high-quality images with an identifiable scleral spur were included, and only the nasal quadrant was analyzed. A single glaucoma specialist analyzed the parameters according to the Zhongshan Angle Assessment Program (ZAAP, Guangzhou, China). Multivariate analysis was performed using mixed effects regression correcting for age, gender, and ethnicity.

Results

89 subjects and 110 eyes were included in this study. The mean age of subjects was 74.83 {+/-} 8.69 years old. Most common diagnoses were POAG and glaucoma suspect (23% and 52%, respectively), and 16% of subjects had an LPI. In multivariate analysis of AS-OCT parameters, decreases in IT750, IT2000, ITCM, ICURV, and pupil size were statistically significant (p<0.05).

Conclusions

After phacoemulsification, eyes with glaucoma as well as glaucoma suspect eyes have thinner irises and smaller pupils. This may lead to less iris-mediated aqueous outflow obstruction, providing support for early phacoemulsification glaucoma treatment.

Translational relevance

Our AS-OCT imaging findings may guide clinical practice as iris parameters become increasingly relevant in preoperative phaco planning.

]]>
<![CDATA[Δ9-Tetrahydrocannabinol and Cannabidiol Differentially Regulate Intraocular Pressure]]> https://www.researchpad.co/article/5c2a78bbd5eed0c48422b890

Purpose

It has been known for nearly 50 years that cannabis and the psychoactive constituent Δ9-tetrahydrocannabinol (THC) reduce intraocular pressure (IOP). Elevated IOP remains the chief hallmark and therapeutic target for glaucoma, a major cause of blindness. THC likely acts via one of the known cannabinoid-related receptors (CB1, CB2, GPR18, GPR119, GPR55) but this has never been determined explicitly. Cannabidiol (CBD) is a second major constituent of cannabis that has been found to be without effect on IOP in most studies.

Methods

Effects of topically applied THC and CBD were tested in living mice by using tonometry and measurements of mRNA levels. In addition the lipidomic consequences of CBD treatment were tested by using lipid analysis.

Results

We now report that a single topical application of THC lowered IOP substantially (∼28%) for 8 hours in male mice. This effect is due to combined activation of CB1 and GPR18 receptors each of which has been shown to lower ocular pressure when activated. We also found that the effect was sex-dependent, being stronger in male mice, and that mRNA levels of CB1 and GPR18 were higher in males. Far from inactive, CBD was found to have two opposing effects on ocular pressure, one of which involved antagonism of tonic signaling. CBD prevents THC from lowering ocular pressure.

Conclusions

We conclude that THC lowers IOP by activating two receptors—CB1 and GPR18—but in a sex-dependent manner. CBD, contrary to expectation, has two opposing effects on IOP and can interfere with the effects of THC.

]]>
<![CDATA[Is Sjögren’s syndrome dry eye similar to dry eye caused by other etiologies? Discriminating different diseases by dry eye tests]]> https://www.researchpad.co/article/5c0ed78bd5eed0c484f1436c

Purpose

Dry Eye Disease (DED) is part of several conditions, including Sjögren’s syndrome (SS) and no single test to diagnosis DED. The present study intends to evaluate whether a set of signs and symptoms of DED can distinguish: a) SS from other non-overlapping systemic diseases related to DED; b) primary and secondary SS.

Methods

182 consecutive patients with DED were evaluated under five groups: SS, graft-versus-host disease (GVHD), Graves' orbitopathy (GO), diabetes mellitus (DM), glaucoma under treatment with benzalkonium chloride medications (BAK). Twenty-four healthy subjects were included as control group (CG). The evaluation consisted of Ocular Surface Disease Index (OSDI), Schirmer test (ST), corneal fluorescein staining (CFS) and tear film break up time (TFBUT). Indeed, a subset of DED patients (n = 130), classified as SS1, SS2 and nonSS (NSS) by the American-European Criteria were compared. Quadratic discriminant analysis (QDA) classified the individuals based on variables collected. The area under Receiver Operating Characteristics (ROC) curve evaluated the classification performance in both comparisons.

Results

Comparing SS with other diseases, QDA showed that the most important variable for classification was OSDI, followed by TFBUT and CFS. Combined, these variables were able to correctly classify 62.6% of subjects in their actual group. At the discretion of the area under the ROC curve, the group with better classification was the control (97.2%), followed by DM (95.5%) and SS (92.5%). DED tests were different among the NSS, SS1 and SS2 groups. The analysis revealed that the combined tests correctly classified 54.6% of the patients in their groups. The area under the ROC curve better classified NSS (79.5%), followed by SS2 (74.4%) and SS1 (69.4%).

Conclusions

Diseases that causes DED, and also SS1, SS2 and NSS are distinguishable conditions, however a single ocular tools was not able to detect the differences among the respective groups.

]]>
<![CDATA[Automated algorithms combining structure and function outperform general ophthalmologists in diagnosing glaucoma]]> https://www.researchpad.co/article/5c117b31d5eed0c484698343

Purpose

To test the ability of machine learning classifiers (MLCs) using optical coherence tomography (OCT) and standard automated perimetry (SAP) parameters to discriminate between healthy and glaucomatous individuals, and to compare it to the diagnostic ability of the combined structure-function index (CSFI), general ophthalmologists and glaucoma specialists.

Design

Cross-sectional prospective study.

Methods

Fifty eight eyes of 58 patients with early to moderate glaucoma (median value of the mean deviation = −3.44 dB; interquartile range, -6.0 to -2.4 dB) and 66 eyes of 66 healthy individuals underwent OCT and SAP tests. The diagnostic accuracy (area under the ROC curve—AUC) of 10 MLCs was compared to those obtained with the CSFI, 3 general ophthalmologists and 3 glaucoma specialists exposed to the same OCT and SAP data.

Results

The AUCs obtained with MLCs ranged from 0.805 (Classification Tree) to 0.931 (Radial Basis Function Network, RBF). The sensitivity at 90% specificity ranged from 51.6% (Classification Tree) to 82.8% (Bagging, Multilayer Perceptron and Support Vector Machine Gaussian). The CSFI had a sensitivity of 79.3% at 90% specificity, and the highest AUC (0.948). General ophthalmologists and glaucoma specialists’ grading had sensitivities of 66.2% and 83.8% at 90% specificity, and AUCs of 0.879 and 0.921, respectively. RBF (the best MLC), the CSFI, and glaucoma specialists showed significantly higher AUCs than that obtained by general ophthalmologists (P<0.05). However, there were no significant differences between the AUCs obtained by RBF, the CSFI, and glaucoma specialists (P>0.25).

Conclusion

Our findings suggest that both MLCs and the CSFI can be helpful in clinical practice and effectively improve glaucoma diagnosis in the primary eye care setting, when there is no glaucoma specialist available.

]]>
<![CDATA[A deep learning approach to automatic detection of early glaucoma from visual fields]]> https://www.researchpad.co/article/5c084192d5eed0c484fca162

Purpose

To investigate the suitability of multi-scale spatial information in 30o visual fields (VF), computed from a Convolutional Neural Network (CNN) classifier, for early-glaucoma vs. control discrimination.

Method

Two data sets of VFs acquired with the OCTOPUS 101 G1 program and the Humphrey Field Analyzer 24–2 pattern were subdivided into control and early-glaucomatous groups, and converted into a new image using a novel voronoi representation to train a custom-designed CNN so to discriminate between control and early-glaucomatous eyes. Saliency maps that highlight what regions of the VF are contributing maximally to the classification decision were computed to provide classification justification. Model fitting was cross-validated and average precision (AP) score performances were computed for our method, Mean Defect (MD), square-root of Loss Variance (sLV), their combination (MD+sLV), and a Neural Network (NN) that does not use convolutional features.

Results

CNN achieved the best AP score (0.874±0.095) across all test folds for one data set compared to others (MD = 0.869±0.064, sLV = 0.775±0.137, MD+sLV = 0.839±0.085, NN = 0.843±0.089) and the third best AP score (0.986 ±0.019) on the other one with slight difference from the other methods (MD = 0.986±0.023, sLV = 0.992±0.016, MD+sLV = 0.987±0.017, NN = 0.985±0.017). In general, CNN consistently led to high AP across different data sets. Qualitatively, computed saliency maps appeared to provide clinically relevant information on the CNN decision for individual VFs.

Conclusion

The proposed CNN offers high classification performance for the discrimination of control and early-glaucoma VFs when compared with standard clinical decision measures. The CNN classification, aided by saliency visualization, may support clinicians in the automatic discrimination of early-glaucomatous and normal VFs.

]]>
<![CDATA[Evaluation of optic nerve subarachnoid space in primary open angle glaucoma using ultrasound examination]]> https://www.researchpad.co/article/5c0841f8d5eed0c484fcb518

Objectives

To measure Optic Nerve Subarachnoid Space (ONSAS) in patients with primary open-angle glaucoma (POAG) and controls using A-scan ultrasound and to evaluate the measurement of the ONSAS in relation to age patient and OCT parameters.

Methods

This retrospective study included 53 consecutive eyes of 27 patients with POAG and 64 normal eyes of 32 controls. Both glaucomatous and control groups were divided into 2 subgroups according to age: <60 age (glaucomatous and control group 1) and 61–90 age (glaucomatous and control group 2).

Results

The ONSAS was significantly lower in all glaucomatous eyes (3.54 ± 0.38) versus normal eyes (3.87 ± 0.32) (p = 0.001). Significant reduction of ONSAS was showed in control group 2 (3.63 mm ± 0.37) compared to control group 1 (3.87 mm ± 0.32) (p = 0.014) and between glaucoma group 1 (3.54 mm ± 0.38) and control group 1 (p = 0.001). While no significant differences were observed between glaucomatous group 2 (3.48 mm ± 0.41) and control group 2 (p = 0.17) and between glaucoma group 1 and glaucoma group 2 (p = 0.609). Lastly, the ONSAS was not significantly associated with GCC and RNFL parameters except for Focal Loss Volume (FLV), Superior RNFL and ONSAS in glaucoma group 1 and for FLV and ONSAS in all glaucomatous group.

Conclusion

Standardized A-scan ultrasound is a non invasive imaging technique with which it is possible to monitor ONSAS changes in glaucomatous patients. The reduction of ONSAS confirm the importance of the lower orbital CSFP as further risk factor in the progression of glaucoma disease.

]]>
<![CDATA[A deep learning model for the detection of both advanced and early glaucoma using fundus photography]]> https://www.researchpad.co/article/5c06f04ed5eed0c484c6d678

Purpose

To build a deep learning model to diagnose glaucoma using fundus photography.

Design

Cross sectional case study Subjects, Participants and Controls: A total of 1,542 photos (786 normal controls, 467 advanced glaucoma and 289 early glaucoma patients) were obtained by fundus photography.

Method

The whole dataset of 1,542 images were split into 754 training, 324 validation and 464 test datasets. These datasets were used to construct simple logistic classification and convolutional neural network using Tensorflow. The same datasets were used to fine tune pre-trained GoogleNet Inception v3 model.

Results

The simple logistic classification model showed a training accuracy of 82.9%, validation accuracy of 79.9% and test accuracy of 77.2%. Convolutional neural network achieved accuracy and area under the receiver operating characteristic curve (AUROC) of 92.2% and 0.98 on the training data, 88.6% and 0.95 on the validation data, and 87.9% and 0.94 on the test data. Transfer-learned GoogleNet Inception v3 model achieved accuracy and AUROC of 99.7% and 0.99 on training data, 87.7% and 0.95 on validation data, and 84.5% and 0.93 on test data.

Conclusion

Both advanced and early glaucoma could be correctly detected via machine learning, using only fundus photographs. Our new model that is trained using convolutional neural network is more efficient for the diagnosis of early glaucoma than previously published models.

]]>
<![CDATA[Classification of primary angle closure spectrum with hierarchical cluster analysis]]> https://www.researchpad.co/article/5b603635463d7e4090b7ce24

Purpose

To classify subjects with primary angle closure into clusters based on features from anterior segment optical coherence tomography (ASOCT) imaging and to explore how these clusters correspond to disease subtypes, including primary angle closure suspect (PACS), primary angle closure glaucoma(PACG), acute primary angle closure (APAC) and fellow eyes of APAC and reveal the factors that become more predominant in each subtype of angle closure.

Method

A cross-sectional study of 248 eyes of 198 subjects(88 PACS eyes, 53 PACG eyes, 54 APAC eyes and 53 fellow eyes of APAC) that underwent complete examination including gonioscopy, A-scan biometry, and ASOCT. An agglomerative hierarchical clustering method was used to classify eyes based on ASOCT parameters.

Results

Statistical clustering analysis produced three clusters among which the anterior segment parameters were significantly different. Cluster 1(43 eyes) had the smallest anterior chamber depth(ACD) and area, as well as the greatest lens vault (p<0.001 for all). Cluster 2(113 eyes) had the thickest iris at 2000 microns(p = 0.048), and largest iris area(p<0.001), and the deepest ACD (p<0.001). Cluster 3(92 eyes) was characterized by elements of both clusters 1 and 2 and a higher iris curvature(p<0.001). There was a statistically significant difference in the distribution of clusters among subtypes of angle closure eyes(p<0.001). Although the patterns of clusters were similar in PACS and PACG eyes, with the majority of the eyes classified into cluster 2(55%, and 62%, respectively), the highest proportion of APAC and fellow eyes were assigned to clusters 1(44%) and 3 (51%), respectively.

Conclusion

Hierarchical cluster analysis identified three clusters with different features. Predominant anatomical components are different among subtypes of primary angle closure.

]]>
<![CDATA[Retinal Nerve Fiber Layer Features Identified by Unsupervised Machine Learning on Optical Coherence Tomography Scans Predict Glaucoma Progression]]> https://www.researchpad.co/article/5b5a80b7463d7e088e574cb2

Purpose

To apply computational techniques to wide-angle swept-source optical coherence tomography (SS-OCT) images to identify novel, glaucoma-related structural features and improve detection of glaucoma and prediction of future glaucomatous progression.

Methods

Wide-angle SS-OCT, OCT circumpapillary retinal nerve fiber layer (cpRNFL) circle scans spectral-domain (SD)-OCT, standard automated perimetry (SAP), and frequency doubling technology (FDT) visual field tests were completed every 3 months for 2 years from a cohort of 28 healthy participants (56 eyes) and 93 glaucoma participants (179 eyes). RNFL thickness maps were extracted from segmented SS-OCT images and an unsupervised machine learning approach based on principal component analysis (PCA) was used to identify novel structural features. Area under the receiver operating characteristic curve (AUC) was used to assess diagnostic accuracy of RNFL PCA for detecting glaucoma and progression compared to SAP, FDT, and cpRNFL measures.

Results

The RNFL PCA features were significantly associated with mean deviation (MD) in both SAP (R2 = 0.49, P < 0.0001) and FDT visual field testing (R2 = 0.48, P < 0.0001), and with mean circumpapillary RNFL thickness (cpRNFLt) from SD-OCT (R2 = 0.58, P < 0.0001). The identified features outperformed each of these measures in detecting glaucoma with an AUC of 0.95 for RNFL PCA compared to an 0.90 for mean cpRNFLt (P = 0.09), 0.86 for SAP MD (P = 0.034), and 0.83 for FDT MD (P = 0.021). Accuracy in predicting progression was also significantly higher for RNFL PCA compared to SAP MD, FDT MD, and mean cpRNFLt (P = 0.046, P = 0.007, and P = 0.044, respectively).

Conclusions

A computational approach can identify structural features that improve glaucoma detection and progression prediction.

]]>