Fuzzy clustering of 24-2 visual field patterns can detect glaucoma progression.

PURPOSE: To represent 24-2 visual field (VF) losses of individual patients using a hybrid approach of archetypal analysis (AA) and fuzzy c-means (FCM) clustering. METHODS: In this multicenter retrospective study, we classified characteristic patterns of 24-2 VF using AA and decomposed them with FCM clustering. We predicted the change in mean deviation (MD) through supervised machine learning from decomposition coefficient change. In addition, we compared the areas under the receiver operating characteristic curves (AUCs) of the decomposition coefficient slopes to detect VF progression using three criteria: MD slope, Visual Field Index slope, and pointwise linear regression analysis. RESULTS: We identified 16 characteristic patterns (archetypes or ATs) of 24-2 VF from 132,938 VFs of 18,033 participants using AA. The hybrid approach using FCM revealed a lower mean squared error and greater correlation coefficient than the AA single approach for predicting MD change (all P ≤ 0.001). Three of 16 AUCs of the FCM decomposition coefficient slopes outperformed the AA decomposition coefficient slopes in detecting VF progression for all three criteria (AT5, superior altitudinal defect; AT10, double arcuate defect; AT13, total loss) (all P ≤ 0.028). CONCLUSION: A hybrid approach combining AA and FCM to analyze 24-2 VF can visualize VF tests in characteristic patterns and enhance detection of VF progression with lossless decomposition.

Bidirectional gated recurrent unit network model can generate future visual field with variable number of input elements.

PURPOSE: This study aimed to predict future visual field tests using a bidirectional gated recurrent unit (Bi-GRU) and assess its performance based on the number of input visual field tests and the prediction time interval. MATERIALS AND METHODS: This study included patients who underwent visual field tests at least four times at five university hospitals between June 2004 and April 2022. All data were accessed in October 2022 for research purposes. In total, 23,517 eyes with 185,858 visual field tests were used as the training dataset, and 1,053 eyes with 9,459 visual field tests were used as the test dataset. The Bi-GRU architecture was designed to take a variable number of visual field tests, ranging from 3 to 80, as input and predict visual field tests at the desired arbitrary time point. It generated the mean deviation (MD), pattern standard deviation (PSD), Visual Field Index (VFI), and total deviation value (TDV) for 54 test points. To analyze the model performance, the mean absolute error between the actual and predicted values was calculated and analyzed for glaucoma severity, number of input visual field tests, and prediction time interval. RESULTS: The prediction errors of the Bi-GRU model for MD, PSD, VFI, and TDV ranged from 1.20 to 1.68 dB, 0.95 to 1.16 dB, 3.64 to 4.51%, and 2.13 to 2.60 dB, respectively, depending on the number of input visual field tests. Prediction errors tended to increase as the prediction time interval increased; however, the difference was not statistically significant. As the severity of glaucoma worsened, the prediction errors significantly increased. CONCLUSION: In clinical practice, the Bi-GRU model can predict future visual field tests at the desired time points using three or more previous visual field tests.

One-Year Outcomes of Ab Externo XEN45 Gel Stent Implantation with an Open Conjunctiva Approach in Patients with Open-Angle Glaucoma.

PURPOSE: This study aimed to evaluate the 1-year surgical outcomes of XEN45 gel stent implantation with an open conjunctiva approach in patients with open-angle glaucoma (OAG). METHODS: This retrospective cohort study included 19 eyes of 19 patients who underwent XEN45 gel stent implantation with an open conjunctival approach. Surgical success was defined by intraocular pressure (IOP) ≤18 mmHg and one of the following: IOP reduction ≥30% or reduction of two glaucoma medications with final IOP if baseline IOP ≤18 mmHg. The cumulative probability of success rate was analyzed using Kaplan-Meier survival analysis. Cox proportional hazard regression analysis was used to assess prognostic factors for surgical failure. RESULTS: IOP reduced from 32.37 ± 12.08 mmHg preoperatively to 15.14 ± 2.25 mmHg at 1 year (p = 0.001). The number of glaucoma medication reduced from 3.89 ± 0.32 preoperatively to 0.86 ± 1.35 at 1 year (p = 0.001). The success rates were 78.9% at 6 months and 73.7% at 1 year. Eyes with bleb horizontal extent ≥2 clock hours at 1 month postoperatively had a significantly higher success rate (log-rank test, p < 0.001). Greater bleb horizontal extent at 2 weeks and 1 month postoperatively was associated with a lower surgical failure rate (2 weeks: hazard ratio, 0.119; p = 0.024; 1 month: hazard ratio, 0.046; p = 0.007). Bleb needling and additional glaucoma surgeries were necessary in 10 (52.6%) and five eyes (26.3%), respectively. CONCLUSIONS: XEN45 gel stent implantation with the open conjunctiva approach is effective in reducing IOP and glaucoma medication for over 1 year in patients with OAG.

Visual field prediction using a deep bidirectional gated recurrent unit network model.

Although deep learning architecture has been used to process sequential data, only a few studies have explored the usefulness of deep learning algorithms to detect glaucoma progression. Here, we proposed a bidirectional gated recurrent unit (Bi-GRU) algorithm to predict visual field loss. In total, 5413 eyes from 3321 patients were included in the training set, whereas 1272 eyes from 1272 patients were included in the test set. Data from five consecutive visual field examinations were used as input; the sixth visual field examinations were compared with predictions by the Bi-GRU. The performance of Bi-GRU was compared with the performances of conventional linear regression (LR) and long short-term memory (LSTM) algorithms. Overall prediction error was significantly lower for Bi-GRU than for LR and LSTM algorithms. In pointwise prediction, Bi-GRU showed the lowest prediction error among the three models in most test locations. Furthermore, Bi-GRU was the least affected model in terms of worsening reliability indices and glaucoma severity. Accurate prediction of visual field loss using the Bi-GRU algorithm may facilitate decision-making regarding the treatment of patients with glaucoma.

Comparison of the Intrableb Characteristics of Anterior Segment Optical Coherence Tomography Imaging in Trabeculectomy according to Amniotic Membrane Transplantation.

INTRODUCTION: The purpose of this study was to compare the characteristics of filtering bleb with anterior segment optical coherence tomography (AS-OCT) according to amniotic membrane transplantation (AMT). METHODS: One hundred and sixteen eyes of 103 glaucoma patients who underwent trabeculectomy with (AMT group; 85 eyes) or without AMT (control group; 31 eyes) were included. Intrableb parameters were evaluated with AS-OCT. Surgical success was defined as an intraocular pressure (IOP) ≤18 mm Hg and IOP reduction ≥20% without medication at the time of AS-OCT examination. Logistic regression analyses were performed to determine factors associated with IOP control. RESULTS: In the eyes with successful IOP control, the fluid-filled space area, score, and height were greater for the AMT group than the control group (all ps < 0.001), while stripping layer thickness was greater and bleb wall reflectivity was lower for the control group than the AMT group (all ps < 0.001). Surgical success in the AMT group was associated with greater fluid-filled space score, lower bleb wall reflectivity, and microcyst formation (odds ratio [OR] = 8.016, 0.913, and 16.202, respectively, all ps ≤ 0.041). Lower bleb wall reflectivity alone was associated with surgical success in the control group (OR = 0.815, p = 0.019). CONCLUSION: The extent of the fluid-filled space was associated with successful IOP control after trabeculectomy with AMT. Hyporeflective bleb wall was associated with successful IOP control in AMT and control groups.

Bleb analysis using anterior segment optical coherence tomography after trabeculectomy with amniotic membrane transplantation.

INTRODUCTION: Little has been known about the intrableb structures associated with bleb function after trabeculectomy with amniotic membrane transplantation (AMT). The aim of this study is to analyze the characteristics of intrableb structures using anterior segment optical coherence tomography (AS-OCT) after trabeculectomy with AMT. METHODS: A total of 68 eyes of 68 patients with primary open-angle glaucoma who underwent trabeculectomy with AMT were included. Surgical success was defined as intraocular pressure (IOP) ≤ 18 mmHg and IOP reduction of ≥ 20% without medication on AS-OCT examination. Intrableb parameters, including bleb height, bleb wall thickness, striping layer thickness, bleb wall reflectivity, fluid-filled space score, fluid-filled space height, and microcyst formation were evaluated using AS-OCT. Logistic regression analysis was performed to determine factors associated with IOP control. RESULTS: Of the 68 eyes, 56 eyes were assigned to the success group and 12 eyes to the failure group. In the success group, bleb height (P = 0.009), bleb wall thickness (P = 0.001), striping layer thickness (P = 0.001), fluid-filled space score (P = 0.001), and frequency of microcyst formation (P = 0.001) were greater than those in the failure group. Bleb wall reflectivity was higher in the failure group than in the success group (P < 0.001). In the univariate logistic regression analysis, previous cataract surgery was significantly associated with surgical failure (odds ratio = 5.769, P = 0.032). CONCLUSION: A posteriorly extending fluid-filled space, tall bleb with low reflectivity, and thick striping layer were characteristics of successful filtering blebs after trabeculectomy with AMT.

Comparison of Retinal Structural and Neurovascular Changes between Patients with and without Amyloid Pathology.

To evaluate whether an impaired anterior visual pathway (retinal structures with microvasculature) is associated with underlying beta-amyloid (Aß) pathologies in patients with Alzheimer's disease dementia (ADD) and mild cognitive impairment (MCI), we compared retinal structural and vascular factors in each subgroup with positive or negative amyloid biomarkers. Twenty-seven patients with dementia, thirty-five with MCI, and nine with cognitively unimpaired (CU) controls were consecutively recruited. All participants were divided into positive Aß (A+) or negative Aß (A-) pathology based on amyloid positron emission tomography or cerebrospinal fluid Aß. The retinal circumpapillary retinal nerve fiber layer thickness (cpRNFLT), macular ganglion cell/inner plexiform layer thickness (mGC/IPLT), and microcirculation of the macular superficial capillary plexus were measured using optical coherence tomography (OCT) and OCT angiography. One eye of each participant was included in the analysis. Retinal structural and vascular factors significantly decreased in the following order: dementia < MCI < CU controls. The A+ group had significantly lower microcirculation in the para- and peri-foveal temporal regions than did the A-. However, the structural and vascular parameters did not differ between the A+ and A- with dementia. The cpRNFLT was unexpectedly greater in the A+ than in the A- with MCI. mGC/IPLT was lower in the A+ CU than in the A- CU. Our findings suggest that retinal structural changes may occur in the preclinical and early stages of dementia but are not highly specific to AD pathophysiology. In contrast, decreased temporal macula microcirculation may be used as a biomarker for the underlying Aß pathology.

Deep learning approaches to predict 10-2 visual field from wide-field swept-source optical coherence tomography en face images in glaucoma.

Close monitoring of central visual field (VF) defects with 10-2 VF helps prevent blindness in glaucoma. We aimed to develop a deep learning model to predict 10-2 VF from wide-field swept-source optical coherence tomography (SS-OCT) images. Macular ganglion cell/inner plexiform layer thickness maps with either wide-field en face images (en face model) or retinal nerve fiber layer thickness maps (RNFLT model) were extracted, combined, and preprocessed. Inception-ResNet-V2 was trained to predict 10-2 VF from combined images. Estimation performance was evaluated using mean absolute error (MAE) between actual and predicted threshold values, and the two models were compared with different input data. The training dataset comprised paired 10-2 VF and SS-OCT images of 3,025 eyes of 1,612 participants and the test dataset of 337 eyes of 186 participants. Global prediction errors (MAEpoint-wise) were 3.10 and 3.17 dB for the en face and RNFLT models, respectively. The en face model performed better than the RNFLT model in superonasal and inferonasal sectors (P = 0.011 and P = 0.030). Prediction errors were smaller in the inferior versus superior hemifields for both models. The deep learning model effectively predicted 10-2 VF from wide-field SS-OCT images and might help clinicians efficiently individualize the frequency of 10-2 VF in clinical practice.

The Effect of Amniotic Membrane Transplantation on Trabeculectomy in Patients with Pseudoexfoliation Glaucoma.

Purpose: The aim is to evaluate the effect of amniotic membrane transplantation (AMT) on trabeculectomy with mitomycin C in patients with pseudoexfoliation glaucoma (PXG). Methods: This retrospective cohort study included 85 eyes of PXG who underwent trabeculectomy with or without AMT (52/33 eyes in the AMT/control group). Surgical success was defined by these criteria: (1) intraocular pressure (IOP) ≤18 mmHg and IOP reduction ≥20% and (2) IOP ≤15 mmHg and IOP reduction ≥25%. Criteria A and B defined complete success rates as patients who met these criteria without medication, respectively. Criteria C and D defined qualified success rates as patients who met these criteria with medication, respectively. Cumulative probabilities of success were compared using the Kaplan-Meier survival analysis. Cox proportional hazard models were used to evaluate the influence of AMT on surgical success accounting for confounding variables. Results: For the AMT group, compared with the control group, the complete success rates at 12 months for criterion A were 86.5% and 63.6%, respectively (P = 0.017) and for criterion B, 86.4% and 63.6% (P = 0.005). The qualified success rates at 12 months for criterion C were 92.1% and 75.1%, respectively (P = 0.047) and for criterion D, 92.1% and 72.1% (P = 0.021). On multivariable Cox regression analyses, AMT was associated with a lower failure rate on criteria A, B, and D (all P ≤ 0.047). Incidence of avascular bleb was higher in the control group than in the AMT group (7 vs 0 eyes; P = 0.004). Conclusions: In patients with PXG, trabeculectomy with AMT was associated with higher success rates and a lower incidence of avascular bleb compared with conventional trabeculectomy. Research Registration. This retrospective cohort study was registered at the Clinical Trial Registry of Korea (https://cris.nih.go.kr/cris/index/index.do, KCT0007228).

Clinical and genetic analysis of Ser341Pro MYOC variant in a Korean family with primary open angle glaucoma.

AIM: To report the first discovery of Ser341Pro myocilin (MYOC) variant in Korea and analyze its clinical characteristics and genetic significance. METHODS: Ten family members from three generations participated in this study and received the thorough ophthalmologic examination. Focused exome sequencing on a proband was performed to confirm the target mutations (MYOC c.1021T>C) in the family members, and the direct sequencing was conducted. Variant was analyzed according to the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG/AMP) guidelines. RESULTS: A nucleotide change from thymine to cytosine at c.1021T>C was found in eight family members. Three members diagnosed with primary open angle glaucoma (POAG) were characterized by severe clinical presentations, high intraocular pressure, and poor response to medical treatment (100% of the patient required filtering surgery). On variant analysis by ACMG/AMP guidelines, Ser341Pro is not found in normal population. Multiple computational predictive programs support a deleterious effect of Ser341Pro variant (PolyPhen 2, SIFT, Mutation Taster). Ser341Pro could be involved in moderate (PM) and supporting (PP) criteria (PM1, PM2, PP2, PP3). Combining the criteria, Ser341Pro has a combination of 2 moderate (PM1+PM2) and 2 supporting (PP2+PP3) criteria, which is interpreted to "likely pathogenic". CONCLUSION: The Ser341Pro variant is correlated with severe phenotype of POAG. There are similar clinical aspects to previous studies: autosomal dominant inheritance, incomplete penetrance (62.50% and 66.67%), and proportion of patients requiring trabeculectomy (100% in both study). According to ACMG/AMP guidelines and the previous basic researches, the Ser341Pro variant had a "strong evidence of pathogenicity (PS3)" and then it could be interpreted to "pathogenic (PS3, PM1, PM2, PP2, PP3)". Additionally, Ser341Pro variant can be reported as "c.1021T>C (p.Ser341Pro), likely pathogenic, POAG, autosomal dominant" according to guideline.

Hybrid dimensionality reduction method based on support vector machine and independent component analysis.

This paper presents a new hybrid dimensionality reduction method to seek projection through optimization of both structural risk (supervised criterion) and data independence (unsupervised criterion). Classification accuracy is used as a metric to evaluate the performance of the method. By minimizing the structural risk, projection originated from the decision boundaries directly improves the classification performance from a supervised perspective. From an unsupervised perspective, projection can also be obtained based on maximum independence among features (or attributes) in data to indirectly achieve better classification accuracy over more intrinsic representation of the data. Orthogonality interrelates the two sets of projections such that minimum redundancy exists between the projections, leading to more effective dimensionality reduction. Experimental results show that the proposed hybrid dimensionality reduction method that satisfies both criteria simultaneously provides higher classification performance, especially for noisy data sets, in relatively lower dimensional space than various existing methods.