Integration and fusion of standard automated perimetry and optical coherence tomography data for improved automated glaucoma diagnostics

Diagnosis of early glaucoma likely combined with high myopia by integrating OCT thickness map and standard automated and Pulsar perimetries

Early-stage glaucoma diagnosis is crucial for preventing permanent structural damage and irreversible vision loss. While various machine-learning approaches have been developed for glaucoma diagnosis, only a few specifically address early-stage detection. Moreover, existing early-stage detection methods rely on unimodal information and exclude subjects with high myopia, which contradicts clinical practice and overlooks the adverse effect of high myopia on prediction performance. To develop a clinically practical tool, this study proposes a deep-learning-based, end-to-end early-stage glaucoma detection framework designed for a cohort likely with high myopia. This framework uniquely integrates functional information from visual field (VF) parameters of standard automated perimetry (SAP) and Pulsar perimetry (PP) with structural information derived from optical coherence tomography (OCT) thickness maps. It comprises three key components: 3D OCT ganglion cell complex (GCC) layer segmentation, thickness map generation, and early-stage glaucoma detection. Evaluated on 394 subjects using five-time, 10-fold cross-validation, the proposed system achieved a mean area under the receiver operating characteristic (ROC) curve of 0.887 ± 0.006, outperforming the Asaoka method without transfer learning and nine models based solely on VF parameters. Results further confirmed that incorporating SAP and PP parameters was essential for mitigating the adverse effects of high myopia.

Fusion of Middle Ear Optical Coherence Tomography and Computed Tomography

Importance

Middle ear optical coherence tomography (OCT) imaging in patients has not previously been directly compared with a standard of care clinical 3-dimensional imaging technology, such as computed tomography (CT).

Objective

To qualitatively compare the capabilities of middle ear OCT with CT in normal and pathological ears on representative slices in coregistered OCT and CT datasets.

Design, Setting, and Participants

This case series included 3 patients and 3 ears: 1 normal middle ear, 1 ear affected by traumatic injury, and 1 ear with cholesteatoma. The ears were imaged with both OCT and high-resolution clinical temporal bone CT. Participants were drawn from the patient population of a tertiary otology clinic. CT and OCT images were aligned using rigid coregistration with manual landmark selection. Data were collected from January 2022 to April 2023, and data were analyzed from February 2022 to December 2023.

Main Outcomes and Measures

Images were analyzed qualitatively for field of view (FOV), resolution, shadowing, artifacts, soft tissue and bony tissue contrast, and presentation of diagnostically important features.

Results

In the 3 imaged ears, OCT was capable of visualizing many of the important features indicative of middle ear pathology. Compared with CT, OCT exhibited a limited FOV largely confined to the mesotympanum and subject to shadowing from bony structures. However, OCT could resolve soft tissue features that were not readily apparent in the CT images to have a higher resolution than CT and to provide excellent anatomical fidelity with CT, which allowed OCT images to be accurately coregistered with CT images.

Conclusions and Relevance

In this case series, while OCT was not capable of replacing CT due to its limited FOV and inability to image through thick bony tissues, it visualized signs of pathology, including some soft tissue features, that are difficult to visualize with CT. Given OCT's ability to image in real time, its compatibility with in-office imaging, and its lack of ionizing radiation, it may, despite its limitations compared with CT, be an appealing imaging modality for many applications in middle ear diagnostics.

Diagnosing glaucoma in primary eye care and the role of Artificial Intelligence applications for reducing the prevalence of undetected glaucoma in Australia

Glaucoma is the commonest cause of irreversible blindness worldwide, with over 70% of people affected remaining undiagnosed. Early detection is crucial for halting progressive visual impairment in glaucoma patients, as there is no cure available. This narrative review aims to: identify reasons for the significant under-diagnosis of glaucoma globally, particularly in Australia, elucidate the role of primary healthcare in glaucoma diagnosis using Australian healthcare as an example, and discuss how recent advances in artificial intelligence (AI) can be implemented to improve diagnostic outcomes. Glaucoma is a prevalent disease in ageing populations and can have improved visual outcomes through appropriate treatment, making it essential for general medical practice. In countries such as Australia, New Zealand, Canada, USA, and the UK, optometrists serve as the gatekeepers for primary eye care, and glaucoma detection often falls on their shoulders. However, there is significant variation in the capacity for glaucoma diagnosis among eye professionals. Automation with Artificial Intelligence (AI) analysis of optic nerve photos can help optometrists identify high-risk changes and mitigate the challenges of image interpretation rapidly and consistently. Despite its potential, there are significant barriers and challenges to address before AI can be deployed in primary healthcare settings, including external validation, high quality real-world implementation, protection of privacy and cybersecurity, and medico-legal implications. Overall, the incorporation of AI technology in primary healthcare has the potential to reduce the global prevalence of undiagnosed glaucoma cases by improving diagnostic accuracy and efficiency.

Screening Strategies and Methodologies

PRCIS

While glaucoma is a leading cause of irreversible vision loss, it presents technical challenges in the design and implementation of screening. New technologies such as PRS and AI offer potential improvements in our ability to identify people at high risk of sight loss from glaucoma and may improve the viability of screening for this important disease.

PURPOSE

To review the current evidence and concepts around screening for glaucoma.

METHODS/RESULTS

A group of glaucoma-focused clinician scientists drew on knowledge and experience around glaucoma, its etiology, and the options for screening. Glaucoma is a chronic progressive optic neuropathy affecting around 76 million individuals worldwide and is the leading cause of irreversible blindness globally. Early stages of the disease are asymptomatic meaning a substantial proportion of cases remain undiagnosed. Early detection and timely intervention reduce the risk of glaucoma-related visual morbidity. However, imperfect tests and a relatively low prevalence currently limit the viability of population-based screening approaches. The diagnostic yield of opportunistic screening strategies, relying on the identification of disease during unrelated health care encounters, such as cataract clinics and diabetic retinopathy screening programs, focusing on older people and/or those with a family history, are hindered by a large number of false-positive and false-negative results. Polygenic risk scores (PRS) offer personalized risk assessment for adult-onset glaucoma. In addition, artificial intelligence (AI) algorithms have shown impressive performance, comparable to expert humans, in discriminating between potentially glaucomatous and non-glaucomatous eyes. These emerging technologies may offer a meaningful improvement in diagnostic yield in glaucoma screening.

CONCLUSIONS

While glaucoma is a leading cause of irreversible vision loss, it presents technical challenges in the design and implementation of screening. New technologies such as PRS and AI offer potential improvements in our ability to identify people at high risk of sight loss from glaucoma and may improve the viability of screening for this important disease.

Using Fused Data from Perimetry and Optical Coherence Tomography to Improve the Detection of Visual Field Progression in Glaucoma

Perimetry and optical coherence tomography (OCT) are both used to monitor glaucoma progression. However, combining these modalities can be a challenge due to differences in data types. To overcome this, we have developed an autoencoder data fusion (AEDF) model to learn compact encoding (AE-fused data) from both perimetry and OCT. The AEDF model, optimized specifically for visual field (VF) progression detection, incorporates an encoding loss to ensure the interpretation of the AE-fused data is similar to VF data while capturing key features from OCT measurements. For model training and evaluation, our study included 2504 longitudinal VF and OCT tests from 140 glaucoma patients. VF progression was determined from linear regression slopes of longitudinal mean deviations. Progression detection with AE-fused data was compared to VF-only data (standard clinical method) as well as data from a Bayesian linear regression (BLR) model. In the initial 2-year follow-up period, AE-fused data achieved a detection F1 score of 0.60 (95% CI: 0.57 to 0.62), significantly outperforming (p < 0.001) the clinical method (0.45, 95% CI: 0.43 to 0.47) and the BLR model (0.48, 95% CI: 0.45 to 0.51). The capacity of the AEDF model to generate clinically interpretable fused data that improves VF progression detection makes it a promising data integration tool in glaucoma management.

Structure-function relationship in glaucoma: Optical coherence tomography en face imaging vs. red-free fundus photography

BACKGROUND/OBJECTIVES

To analyse retinal nerve fibre layer (RNFL) defect measurements obtained from red-free fundus photography and optical coherence tomography (OCT) en face imaging, respectively, and to compare them for the strength of the structure-function association.

SUBJECTS/METHODS

Two hundred and fifty-six glaucomatous eyes of 256 patients with localized RNFL defect on red-free fundus photography were enrolled. A subgroup analysis included 81 highly myopic eyes (≤ -6.0 dioptres). Angular width of RNFL defect was compared between red-free fundus photography (i.e., red-free RNFL defect) and OCT en face imaging (i.e., en face RNFL defect). The correlation between angular width of each RNFL defect and functional outcomes, reported as mean deviation (MD) and pattern standard deviation (PSD), were assessed and compared.

RESULTS

The angular width of en face RNFL defect was measured smaller than that of red-free RNFL defect in 91.0% eyes (mean difference, 19.98°). The association of en face RNFL defect with MD and PSD was stronger (R2 = 0.311 and R2 = 0.372, respectively) than that of red-free RNFL defect with MD and PSD (R2 = 0.162 and R2 = 0.137, respectively) (P < 0.05 for all). Especially in highly myopic eyes, the association of en face RNFL defect with MD and PSD was much stronger (R2 = 0.503 and R2 = 0.555, respectively) than that of red-free RNFL defect with MD and PSD (R2 = 0.216 and R2 = 0.166, respectively) (P < 0.05 for all).

CONCLUSIONS

En face RNFL defect showed a higher correlation with severity of visual field loss than did red-free RNFL defect. The same dynamic was observed for highly myopic eyes.

Spatial Cluster Patterns of Retinal Sensitivity Loss in Intermediate Age-Related Macular Degeneration Features

Purpose

To examine spatial patterns of retinal sensitivity loss in the three key features of intermediate age-related macular degeneration (iAMD).

Methods

One-hundred individuals (53 iAMD, 47 normal) underwent 10-2 mesopic microperimetry testing in one eye. Pointwise sensitivities (dB) were corrected for age, sex, iAMD status, and co-presence of co-localized key iAMD features: drusen load, pigmentary abnormalities, and reticular pseudodrusen (RPD). Clusters (labeled by ranks of magnitude C-2, C-1, C0) were derived from pointwise sensitivities and then assessed by quadrants and eccentricity/rings.

Results

Two clusters of decreased sensitivities were evident in iAMD versus normal: C-2, -1.67 dB (95% CI (confidence intervals), -2.36 to -0.98; P < 0.0001); C-1, -0.93 dB (95% CI, -1.5 to -0.36; P < 0.01). One cluster of decreased sensitivity was independently associated each with increased drusen load (13.57 µm increase per -1 dB; P < 0.0001), pigmentary abnormalities (C-1: -2.23 dB; 95% CI, -3.36 to -1.1; P < 0.01), and RPD (C-1: -1.07 dB; 95% CI, -2 to -0.14; P < 0.01). Sensitivity loss in iAMD was biased toward the superior and central macula (P = 0.16 to <0.0001), aligning with structural distributions of features. However, sensitivity loss associated with drusen load also extended to the peripheral macula (P < 0.0001) with paracentral sparing, which was discordant with the central distribution of drusen.

Conclusions

Drusen load, pigmentary abnormalities, and RPD are associated with patterns of retinal sensitivity loss commonly demonstrating superior and central bias. Results highlighted that a clinical focus on these three key iAMD features using structural measures alone does not capture the complex, spatial extent of vision-related functional impairment in iAMD.

Translational Relevance

Defining the spatial patterns of retinal sensitivity loss in iAMD can facilitate a targeted visual field protocol for iAMD assessment.

Visual fields in glaucoma: Where are we now?

Visual fields are an integral part of glaucoma diagnosis and management. COVID has heightened the awareness of the potential for viral spread with the practice of visual fields modified. Mask artefacts can occur due to fogging of the inferior rim of the trail lens. Fortunately, the risk of airborne transmission when field testing is low. The 24-2c may be useful to detect early disease and the 10-2 more sensitive to detect advanced loss. The SITA faster test algorithm is able to reduce testing time thereby improving clinic efficiency, however, may show milder results for moderate or severe glaucoma. The technician has an important role of supervising the visual field performance to achieve reliable output. Home monitoring can provide earlier detection of progression and thus improve monitoring of glaucoma as well as reduce the burden of in-clinic assessments. Artificial Intelligence has been found to have high sensitivity and specificity compared to expert observers in detecting field abnormalities and progression as well as integrating structure with function. Although these advances will improve efficiency and guide accuracy, there will remain a need for clinicians to interpret the results and instigate management.

A Multimodal Classification Architecture for the Severity Diagnosis of Glaucoma Based on Deep Learning

Glaucoma is an optic neuropathy that leads to characteristic visual field defects. However, there is no cure for glaucoma, so the diagnosis of its severity is essential for its prevention. In this paper, we propose a multimodal classification architecture based on deep learning for the severity diagnosis of glaucoma. In this architecture, a gray scale image of the visual field is first reconstructed with a higher resolution in the preprocessing stage, and more subtle feature information is provided for glaucoma diagnosis. We then use multimodal fusion technology to integrate fundus images and gray scale images of the visual field as the input of this architecture. Finally, the inherent limitation of convolutional neural networks (CNNs) is addressed by replacing the original classifier with the proposed classifier. Our architecture is trained and tested on the datasets provided by the First Affiliated Hospital of Kunming Medical University, and the results show that the proposed architecture achieves superior performance for glaucoma diagnosis.

[Artificial intelligence and glaucoma: A literature review]

In recent years, research in artificial intelligence (AI) has experienced an unprecedented surge in the field of ophthalmology, in particular glaucoma. The diagnosis and follow-up of glaucoma is complex and relies on a body of clinical evidence and ancillary tests. This large amount of information from structural and functional testing of the optic nerve and macula makes glaucoma a particularly appropriate field for the application of AI. In this paper, we will review work using AI in the field of glaucoma, whether for screening, diagnosis or detection of progression. Many AI strategies have shown promising results for glaucoma detection using fundus photography, optical coherence tomography, or automated perimetry. The combination of these imaging modalities increases the performance of AI algorithms, with results comparable to those of humans. We will discuss potential applications as well as obstacles and limitations to the deployment and validation of such models. While there is no doubt that AI has the potential to revolutionize glaucoma management and screening, research in the coming years will need to address unavoidable questions regarding the clinical significance of such results and the explicability of the predictions.

Computational Audiology: New Approaches to Advance Hearing Health Care in the Digital Age

The global digital transformation enables computational audiology for advanced clinical applications that can reduce the global burden of hearing loss. In this article, we describe emerging hearing-related artificial intelligence applications and argue for their potential to improve access, precision, and efficiency of hearing health care services. Also, we raise awareness of risks that must be addressed to enable a safe digital transformation in audiology. We envision a future where computational audiology is implemented via interoperable systems using shared data and where health care providers adopt expanded roles within a network of distributed expertise. This effort should take place in a health care system where privacy, responsibility of each stakeholder, and patients' safety and autonomy are all guarded by design.

Deep Relation Transformer for Diagnosing Glaucoma With Optical Coherence Tomography and Visual Field Function

Glaucoma is the leading reason for irreversible blindness. Early detection and timely treatment of glaucoma are essential for preventing visual field loss or even blindness. In clinical practice, Optical Coherence Tomography (OCT) and Visual Field (VF) exams are two widely-used and complementary techniques for diagnosing glaucoma. OCT provides quantitative measurements of the optic nerve head (ONH) structure, while VF test is the functional assessment of peripheral vision. In this paper, we propose a Deep Relation Transformer (DRT) to perform glaucoma diagnosis with OCT and VF information combined. A novel deep reasoning mechanism is proposed to explore implicit pairwise relations between OCT and VF information in global and regional manners. With the pairwise relations, a carefully-designed deep transformer mechanism is developed to enhance the representation with complementary information for each modal. Based on reasoning and transformer mechanisms, three successive modules are designed to extract and collect valuable information for glaucoma diagnosis, the global relation module, the guided regional relation module, and the interaction transformer module, namely. Moreover, we build a large dataset, namely ZOC-OCT&VF dataset, which includes 1395 OCT-VF pairs for developing and evaluating our DRT. We conduct extensive experiments to validate the effectiveness of the proposed method. Experimental results show that our method achieves 88.3% accuracy and outperforms the existing single-modal approaches with a large margin. The codes and dataset will be publicly available in the future.

Prediction of Visual Field Progression from OCT Structural Measures in Moderate to Advanced Glaucoma

PURPOSE

To test the hypothesis that visual field (VF) progression can be predicted from baseline and longitudinal optical coherence tomography (OCT) structural measurements.

DESIGN

Prospective cohort study.

METHODS

A total of 104 eyes (104 patients) with ≥3 years of follow-up and ≥5 VF examinations were enrolled. We defined VF progression based on pointwise linear regression on 24-2 VF (≥3 locations with slope less than or equal to -1.0 dB/year and P < .01). We used elastic net logistic regression (ENR) and machine learning to predict VF progression with demographics, baseline circumpapillary retinal nerve fiber layer (RNFL), macular ganglion cell/inner plexiform layer (GCIPL) thickness, and RNFL and GCIPL change rates at central 24 superpixels and 3 eccentricities, 3.4°, 5.5°, and 6.8°, from fovea and hemimaculas. Areas-under-ROC curves (AUC) were used to compare models.

RESULTS

Average ± SD follow-up and VF examinations were 4.5 ± 0.9 years and 8.7 ± 1.6, respectively. VF progression was detected in 23 eyes (22%). ENR selected rates of change of superotemporal RNFL sector and GCIPL change rates in 5 central superpixels and at 3.4° and 5.6° eccentricities as the best predictor subset (AUC = 0.79 ± 0.12). Best machine learning predictors consisted of baseline superior hemimacular GCIPL thickness and GCIPL change rates at 3.4° eccentricity and 3 central superpixels (AUC = 0.81 ± 0.10). Models using GCIPL-only structural variables performed better than RNFL-only models.

CONCLUSIONS

VF progression can be predicted with clinically relevant accuracy from baseline and longitudinal structural data. Further refinement of proposed models would assist clinicians with timely prediction of functional glaucoma progression and clinical decision making.

Deep Learning Ensemble Method for Classifying Glaucoma Stages Using Fundus Photographs and Convolutional Neural Networks

Purpose: This study developed and evaluated a deep learning ensemble method to automatically grade the stages of glaucoma depending on its severity.Materials and Methods: After cross-validation of three glaucoma specialists, the final dataset comprised of 3,460 fundus photographs taken from 2,204 patients were divided into three classes: unaffected controls, early-stage glaucoma, and late-stage glaucoma. The mean deviation value of standard automated perimetry was used to classify the glaucoma cases. We modeled 56 convolutional neural networks (CNN) with different characteristics and developed an ensemble system to derive the best performance by combining several modeling results.Results: The proposed method with an accuracy of 88.1% and an average area under the receiver operating characteristic of 0.975 demonstrates significantly better performance to classify glaucoma stages compared to the best single CNN model that has an accuracy of 85.2% and an average area under the receiver operating characteristic of 0.950. The false negative is the least adjacent misprediction, and it is less in the proposed method than in the best single CNN model.Conclusions: The method of averaging multiple CNN models can better classify glaucoma stages by using fundus photographs than a single CNN model. The ensemble method would be useful as a clinical decision support system in glaucoma screening for primary care because it provides high and stable performance with a relatively small amount of data.

Artificial Intelligence Algorithms to Diagnose Glaucoma and Detect Glaucoma Progression: Translation to Clinical Practice

Purpose

This concise review aims to explore the potential for the clinical implementation of artificial intelligence (AI) strategies for detecting glaucoma and monitoring glaucoma progression.

Methods

Nonsystematic literature review using the search combinations “Artificial Intelligence,” “Deep Learning,” “Machine Learning,” “Neural Networks,” “Bayesian Networks,” “Glaucoma Diagnosis,” and “Glaucoma Progression.” Information on sensitivity and specificity regarding glaucoma diagnosis and progression analysis as well as methodological details were extracted.

Results

Numerous AI strategies provide promising levels of specificity and sensitivity for structural (e.g. optical coherence tomography [OCT] imaging, fundus photography) and functional (visual field [VF] testing) test modalities used for the detection of glaucoma. Area under receiver operating curve (AROC) values of > 0.90 were achieved with every modality. Combining structural and functional inputs has been shown to even more improve the diagnostic ability. Regarding glaucoma progression, AI strategies can detect progression earlier than conventional methods or potentially from one single VF test.

Conclusions

AI algorithms applied to fundus photographs for screening purposes may provide good results using a simple and widely accessible test. However, for patients who are likely to have glaucoma more sophisticated methods should be used including data from OCT and perimetry. Outputs may serve as an adjunct to assist clinical decision making, whereas also enhancing the efficiency, productivity, and quality of the delivery of glaucoma care. Patients with diagnosed glaucoma may benefit from future algorithms to evaluate their risk of progression. Challenges are yet to be overcome, including the external validity of AI strategies, a move from a “black box” toward “explainable AI,” and likely regulatory hurdles. However, it is clear that AI can enhance the role of specialist clinicians and will inevitably shape the future of the delivery of glaucoma care to the next generation.

Translational Relevance

The promising levels of diagnostic accuracy reported by AI strategies across the modalities used in clinical practice for glaucoma detection can pave the way for the development of reliable models appropriate for their translation into clinical practice. Future incorporation of AI into healthcare models may help address the current limitations of access and timely management of patients with glaucoma across the world.

Combining optical coherence tomography with visual field data to rapidly detect disease progression in glaucoma: a diagnostic accuracy study

BACKGROUND

Progressive optic nerve damage in glaucoma results in vision loss, quantifiable with visual field (VF) testing. VF measurements are, however, highly variable, making identification of worsening vision ('progression') challenging. Glaucomatous optic nerve damage can also be measured with imaging techniques such as optical coherence tomography (OCT).

OBJECTIVE

To compare statistical methods that combine VF and OCT data with VF-only methods to establish whether or not these allow (1) more rapid identification of glaucoma progression and (2) shorter or smaller clinical trials.

DESIGN

Method 'hit rate' (related to sensitivity) was evaluated in subsets of the United Kingdom Glaucoma Treatment Study (UKGTS) and specificity was evaluated in 72 stable glaucoma patients who had 11 VF and OCT tests within 3 months (the RAPID data set). The reference progression detection method was based on Guided Progression Analysis™ (GPA) Software (Carl Zeiss Meditec Inc., Dublin, CA, USA). Index methods were based on previously described approaches [Analysis with Non-Stationary Weibull Error Regression and Spatial enhancement (ANSWERS), Permutation analyses Of Pointwise Linear Regression (PoPLR) and structure-guided ANSWERS (sANSWERS)] or newly developed methods based on Permutation Test (PERM), multivariate hierarchical models with multiple imputation for censored values (MaHMIC) and multivariate generalised estimating equations with multiple imputation for censored values (MaGIC).

SETTING

Ten university and general ophthalmology units (UKGTS) and a single university ophthalmology unit (RAPID).

PARTICIPANTS

UKGTS participants were newly diagnosed glaucoma patients randomised to intraocular pressure-lowering drops or placebo. RAPID participants had glaucomatous VF loss, were on treatment and were clinically stable.

INTERVENTIONS

24-2 VF tests with the Humphrey Field Analyzer and optic nerve imaging with time-domain (TD) Stratus OCT™ (Carl Zeiss Meditec Inc., Dublin, CA, USA).

MAIN OUTCOME MEASURES

Criterion hit rate and specificity, time to progression, future VF prediction error, proportion progressing in UKGTS treatment groups, hazard ratios (HRs) and study sample size.

RESULTS

Criterion specificity was 95% for all tests; the hit rate was 22.2% for GPA, 41.6% for PoPLR, 53.8% for ANSWERS and 61.3% for sANSWERS (all comparisons p ≤ 0.042). Mean survival time (weeks) was 93.6 for GPA, 82.5 for PoPLR, 72.0 for ANSWERS and 69.1 for sANSWERS. The median prediction errors (decibels) when the initial trend was used to predict the final VF were 3.8 (5th to 95th percentile 1.7 to 7.6) for PoPLR, 3.0 (5th to 95th percentile 1.5 to 5.7) for ANSWERS and 2.3 (5th to 95th percentile 1.3 to 4.5) for sANSWERS. HRs were 0.57 [95% confidence interval (CI) 0.34 to 0.90; p = 0.016] for GPA, 0.59 (95% CI 0.42 to 0.83; p = 0.002) for PoPLR, 0.76 (95% CI 0.56 to 1.02; p = 0.065) for ANSWERS and 0.70 (95% CI 0.53 to 0.93; p = 0.012) for sANSWERS. Sample size estimates were not reduced using methods including OCT data. PERM hit rates were between 8.3% and 17.4%. Treatment effects were non-significant in MaHMIC and MaGIC analyses; statistical significance was altered little by incorporating imaging.

LIMITATIONS

TD OCT is less precise than current imaging technology; current OCT technology would likely perform better. The size of the RAPID data set limited the precision of criterion specificity estimates.

CONCLUSIONS

The sANSWERS method combining VF and OCT data had a higher hit rate and identified progression more quickly than the reference and other VF-only methods, and produced more accurate estimates of the progression rate, but did not increase treatment effect statistical significance. Similar studies with current OCT technology need to be undertaken and the statistical methods need refinement.

TRIAL REGISTRATION

Current Controlled Trials ISRCTN96423140.

FUNDING

This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 22, No. 4. See the NIHR Journals Library website for further project information. Data analysed in the study were from the UKGTS. Funding for the UKGTS was provided through an unrestricted investigator-initiated research grant from Pfizer Inc. (New York, NY, USA), with supplementary funding from the NIHR Biomedical Research Centre at Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology, London, UK. Imaging equipment loans were made by Heidelberg Engineering, Carl Zeiss Meditec and Optovue (Fremont, CA, USA). Pfizer, Heidelberg Engineering, Carl Zeiss Meditec and Optovue had no input into the design, conduct, analysis or reporting of any of the UKGTS findings or this work. The sponsor for both the UKGTS and RAPID data collection was Moorfields Eye Hospital NHS Foundation Trust. David F Garway-Heath, Tuan-Anh Ho and Haogang Zhu are partly funded by the NIHR Biomedical Research Centre based at Moorfields Eye Hospital and UCL Institute of Ophthalmology. David F Garway-Heath's chair at University College London (UCL) is supported by funding from the International Glaucoma Association.

Machine learning models based on the dimensionality reduction of standard automated perimetry data for glaucoma diagnosis

INTRODUCTION

Visual field testing via standard automated perimetry (SAP) is a commonly used glaucoma diagnosis method. Applying machine learning techniques to the visual field test results, a valid clinical diagnosis of glaucoma solely based on the SAP data is provided. In order to reflect structural-functional patterns of glaucoma on the automated diagnostic models, we propose composite variables derived from anatomically grouped visual field clusters to improve the prediction performance. A set of machine learning-based diagnostic models are designed that implement different input data manipulation, dimensionality reduction, and classification methods.

METHODS

Visual field testing data of 375 healthy and 257 glaucomatous eyes were used to build the diagnostic models. Three kinds of composite variables derived from the Garway-Heath map and the glaucoma hemifield test (GHT) sector map were included in the input variables in addition to the 52 SAP visual filed locations. Dimensionality reduction was conducted to select important variables so as to alleviate high-dimensionality problems. To validate the proposed methods, we applied four classifiers-linear discriminant analysis, naïve Bayes classifier, support vector machines, and artificial neural networks-and four dimensionality reduction methods-Pearson correlation coefficient-based variable selection, Markov blanket variable selection, the minimum redundancy maximum relevance algorithm, and principal component analysis- and compared their classification performances.

RESULTS

For all tested combinations, the classification performance improved when the proposed composite variables and dimensionality reduction techniques were implemented. The combination of total deviation values, the GHT sector map, support vector machines, and Markov blanket variable selection obtains the best performance: an area under the receiver operating characteristic curve (AUC) of 0.912.

CONCLUSION

A glaucoma diagnosis model giving an AUC of 0.912 was constructed by applying machine learning techniques to SAP data. The results show that dimensionality reduction not only reduces dimensions of the input space but also enhances the classification performance. The variable selection results show that the proposed composite variables from visual field clustering play a key role in the diagnosis model.

Optical coherence tomography for glaucoma diagnosis: An evidence based meta-analysis

Purpose

Early detection, monitoring and understanding of changes in the retina are central to the diagnosis of glaucomatous optic neuropathy, and vital to reduce visual loss from this progressive condition. The main objective of this investigation was to compare glaucoma diagnostic accuracy of commercially available optical coherence tomography (OCT) devices (Zeiss Stratus, Zeiss Cirrus, Heidelberg Spectralis and Optovue RTVue, and Topcon 3D-OCT).

Patients

16,104 glaucomatous and 11,543 normal eyes reported in 150 studies.

Methods

Between Jan. 2017 and Feb 2017, MEDLINE^®, EMBASE^®, CINAHL^®, Cochrane Library^®, Web of Science^®, and BIOSIS^® were searched for studies assessing glaucoma diagnostic accuracy of the aforementioned OCT devices. Meta-analysis was performed pooling area under the receiver operating characteristic curve (AUROC) estimates for all devices, stratified by OCT type (RNFL, macula), and area imaged.

Results

150 studies with 16,104 glaucomatous and 11,543 normal control eyes were included. Key findings: AUROC of glaucoma diagnosis for RNFL average for all glaucoma patients was 0.897 (0.887–0.906, n = 16,782 patient eyes), for macula ganglion cell complex (GCC) was 0.885 (0.869–0.901, n = 4841 eyes), for macula ganglion cell inner plexiform layer (GCIPL) was 0.858 (0.835–0.880, n = 4211 eyes), and for total macular thickness was 0.795 (0.754–0.834, n = 1063 eyes).

Conclusion

The classification capability was similar across all 5 OCT devices. More diagnostically favorable AUROCs were demonstrated in patients with increased glaucoma severity. Diagnostic accuracy of RNFL and segmented macular regions (GCIPL, GCC) scans were similar and higher than total macular thickness. This study provides a synthesis of contemporary evidence with features of robust inclusion criteria and large sample size. These findings may provide guidance to clinicians when navigating this rapidly evolving diagnostic area characterized by numerous options.

Evaluation of Visual Field and Imaging Outcomes for Glaucoma Clinical Trials (An American Ophthalomological Society Thesis)

PURPOSE

To evaluate the ability of various visual field (VF) analysis methods to discriminate treatment groups in glaucoma clinical trials and establish the value of time-domain optical coherence tomography (TD OCT) imaging as an additional outcome.

METHODS

VFs and retinal nerve fibre layer thickness (RNFLT) measurements (acquired by TD OCT) from 373 glaucoma patients in the UK Glaucoma Treatment Study (UKGTS) at up to 11 scheduled visits over a 2 year interval formed the cohort to assess the sensitivity of progression analysis methods. Specificity was assessed in 78 glaucoma patients with up to 11 repeated VF and OCT RNFLT measurements over a 3 month interval. Growth curve models assessed the difference in VF and RNFLT rate of change between treatment groups. Incident progression was identified by 3 VF-based methods: Guided Progression Analysis (GPA), 'ANSWERS' and 'PoPLR', and one based on VFs and RNFLT: 'sANSWERS'. Sensitivity, specificity and discrimination between treatment groups were evaluated.

RESULTS

The rate of VF change was significantly faster in the placebo, compared to active treatment, group (-0.29 vs +0.03 dB/year, P<.001); the rate of RNFLT change was not different (-1.7 vs -1.1 dB/year, P=.14). After 18 months and at 95% specificity, the sensitivity of ANSWERS and PoPLR was similar (35%); sANSWERS achieved a sensitivity of 70%. GPA, ANSWERS and PoPLR discriminated treatment groups with similar statistical significance; sANSWERS did not discriminate treatment groups.

CONCLUSIONS

Although the VF progression-detection method including VF and RNFLT measurements is more sensitive, it does not improve discrimination between treatment arms.

Effect of head tilt on repeatability of optic nerve head parameters using cirrus spectral-domain optical coherence tomography

AIM

To assess the repeatability of measuring optic nerve head (ONH) parameters using the Cirrus optical coherence tomography (OCT), as well as to assess the effect of head tilt on these measurements.

METHODS

Thirty healthy participants with no evidence of glaucoma were recruited for the study. Visual acuity, intraocular pressure, standard automated perimetry and ocular examination were performed for each participant. One eye was then randomly selected and scanned undilated with the Cirrus OCT in 3 positions (neutral, 30° right tilt and 30° left tilt).

RESULTS

Data collected from 29 eyes were used for analysis. One patient was omitted due to poor scan quality. The repeatability of the ONH parameters was analyzed using analysis of variance, coefficient of variation (COV) and intraclass correlation coefficient (ICC). Analysis of variance showed no statistically significant difference between 3 scans in a single position. There was good agreement between measurements (ICC 0.919-0.996, COV 1.94%-5.48%). Even with the presence of head tilt, repeated scans in the 3 positions showed good agreement as well (ICC 0.888-0.996, COV 2.04%-5.39%).

CONCLUSION

Serial measurements of ONH parameters using the Cirrus OCT are found to have good repeatability. The ONH parameters with Cirrus OCT also maintain good repeatability despite head tilt.

Color Reflectivity Discretization Analysis of OCT Images in the Detection of Glaucomatous Nerve Fiber Layer Defects

PURPOSE

To compare the ability of Cirrus retinal nerve fiber layer (RNFL) thickness and the Color Reflectivity Discretization Analysis (CORDA), a novel optical coherence tomography (OCT) analysis method, to differentiate between normal subjects, glaucoma suspects, and glaucoma patients.

PATIENTS AND METHODS

Analysis of peripapillary OCT images using Cirrus SD-OCT (optic nerve head cube 200 × 200 protocol) and postacquisition CORDA analysis of peripapillary RNFL B-scan images was performed. In total, 291 eyes of 148 subjects (94 normal eyes, 100 primary open-angle glaucoma suspect eyes, and 97 eyes with primary open-angle glaucoma) were included. Area under the receiver operating characteristic curve was estimated for each region and method (Cirrus vs. CORDA) for differentiating eyes with glaucoma, and those that are glaucoma suspect, from normal eyes.

RESULTS

CORDA HR1 parameter discriminated glaucoma patients from normal subjects more accurately than Cirrus RNFL thickness in nasal (P = 0.003) and temporal (P = 0.001) regions. HR1 showed greater area under the receiver operating characteristic curve than Cirrus RNFL thickness when discriminating glaucoma suspects from normal subjects in the superior (P = 0.02), nasal (P = 0.003), and temporal (P = 0.001) regions. Both were similar for mean and the inferior regions.

CONCLUSIONS

In this study, the novel CORDA HR1 differentiated between normal subjects and glaucoma suspects more accurately than Cirrus RNFL, and in temporal and nasal regions when discriminating between normal and glaucomatous eyes. CORDA analysis may improve the diagnostic accuracy of Cirrus OCT for glaucoma and glaucoma suspects.

A survey on computer aided diagnosis for ocular diseases

Background

Computer Aided Diagnosis (CAD), which can automate the detection process for ocular diseases, has attracted extensive attention from clinicians and researchers alike. It not only alleviates the burden on the clinicians by providing objective opinion with valuable insights, but also offers early detection and easy access for patients.

Method

We review ocular CAD methodologies for various data types. For each data type, we investigate the databases and the algorithms to detect different ocular diseases. Their advantages and shortcomings are analyzed and discussed.

Result

We have studied three types of data (i.e., clinical, genetic and imaging) that have been commonly used in existing methods for CAD. The recent developments in methods used in CAD of ocular diseases (such as Diabetic Retinopathy, Glaucoma, Age-related Macular Degeneration and Pathological Myopia) are investigated and summarized comprehensively.

Conclusion

While CAD for ocular diseases has shown considerable progress over the past years, the clinical importance of fully automatic CAD systems which are able to embed clinical knowledge and integrate heterogeneous data sources still show great potential for future breakthrough.

On improving the use of OCT imaging for detecting glaucomatous damage

AIMS

To describe two approaches for improving the detection of glaucomatous damage seen with optical coherence tomography (OCT).

METHODS

The two approaches described were: one, a visual analysis of the high-quality OCT circle scans and two, a comparison of local visual field sensitivity loss to local OCT retinal ganglion cell plus inner plexiform (RGC+) and retinal nerve fibre layer (RNFL) thinning. OCT images were obtained from glaucoma patients and suspects using a spectral domain OCT machine and commercially available scanning protocols. A high-quality peripapillary circle scan (average of 50), a three-dimensional (3D) scan of the optic disc, and a 3D scan of the macula were obtained. RGC+ and RNFL thickness and probability plots were generated from the 3D scans.

RESULTS

A close visual analysis of a high-quality circle scan can help avoid both false positive and false negative errors. Similarly, to avoid these errors, the location of abnormal visual field points should be compared to regions of abnormal RGC+ and RNFL thickness.

CONCLUSIONS

To improve the sensitivity and specificity of OCT imaging, high-quality images should be visually scrutinised and topographical information from visual fields and OCT scans combined.

Improving glaucoma detection using spatially correspondent clusters of damage and by combining standard automated perimetry and optical coherence tomography

PURPOSE

To improve the detection of glaucoma, techniques for assessing local patterns of damage and for combining structure and function were developed.

METHODS

Standard automated perimetry (SAP) and frequency-domain optical coherence tomography (fdOCT) data, consisting of macular retinal ganglion cell plus inner plexiform layer (mRGCPL) as well as macular and optic disc retinal nerve fiber layer (mRNFL and dRNFL) thicknesses, were collected from 52 eyes of 52 healthy controls and 156 eyes of 96 glaucoma suspects and patients. In addition to generating simple global metrics, SAP and fdOCT data were searched for contiguous clusters of abnormal points and converted to a continuous metric (pcc). The pcc metric, along with simpler methods, was used to combine the information from the SAP and fdOCT. The performance of different methods was assessed using the area under receiver operator characteristic curves (AROC scores).

RESULTS

The pcc metric performed better than simple global measures for both the fdOCT and SAP. The best combined structure-function metric (mRGCPL&SAP pcc, AROC = 0.868 ± 0.032) was better (statistically significant) than the best metrics for independent measures of structure and function. When SAP was used as part of the inclusion and exclusion criteria, AROC scores increased for all metrics, including the best combined structure-function metric (AROC = 0.975 ± 0.014).

CONCLUSIONS

A combined structure-function metric improved the detection of glaucomatous eyes. Overall, the primary sources of value-added for glaucoma detection stem from the continuous cluster search (the pcc), the mRGCPL data, and the combination of structure and function.

Choice of statistical method influences apparent association between structure and function in glaucoma

PURPOSE

The aim of this study was to explore how different statistical methods may lead to inconsistent inferences about the association between structure and function in glaucoma.

METHODS

Two datasets from published studies were selected for their illustrative value. The first consisted of measurements of neuroretinal rim area in the superior-temporal sector paired with the corresponding visual field sensitivity. The second consisted of measurements of average retinal nerve fiber layer thickness over all sectors paired with the corresponding visual field sensitivity. Statistical methods included linear and segmented regression, and a nonparametric local-linear fit known as loess. The analyses were repeated with all measurements expressed as percent of mean normal.

RESULTS

Slopes from linear fits to the data changed by a factor of 10 depending on the linear regression method applied. Inferences about whether structural abnormality precedes functional abnormality varied with the statistical design and the units of measure used.

CONCLUSIONS

The apparent association between structure and function in glaucoma, and consequent interpretation, varies with the statistical method and units of measure. Awareness of the limitations of any statistical analysis is necessary to avoid finding spurious results that ultimately may lead to inadequate clinical recommendations.

Imaging of retinal ganglion cells in glaucoma: pitfalls and challenges

Imaging has gained a key role in modern glaucoma management. Traditionally, interest was directed toward the appearance of the optic nerve head and the retinal nerve fiber layer. With the improvement of the resolution of optical coherence tomography, the ganglion cell complex has also become routinely accessible in the clinic. Further advances have been made in understanding the structure-function relationship in glaucoma. Nevertheless, direct imaging of the retinal ganglion cells in glaucoma would be advantageous. With the currently used techniques, this goal cannot be achieved, because the transversal resolution is limited by aberrations of the eye. The use of adaptive optics has significantly improved transversal resolution, and the imaging of several cell types including cones and astrocytes has become possible. Imaging of retinal ganglion cells, however, still remains a problem, because of the transparency of these cells. However, the visualization of retinal ganglion cells and their dendrites has been achieved in animal models. Furthermore, attempts have been made to visualize the apoptosis of retinal ganglion cells in vivo. Implementation of these techniques in clinical practice will probably improve glaucoma care and facilitate the development of neuroprotective strategies.

'Structure-function relationship' in glaucoma: past thinking and current concepts

An understanding of the relationship between functional and structural measures in primary open-angle glaucoma is necessary for both grading the severity of disease and for understanding the natural history of the condition. This article outlines the current evidence for the nature of this relationship and highlights the current mathematical models linking structure and function. Large clinical trials demonstrate that both structural and functional change are apparent in advanced stages of disease, and at an individual level, detectable structural abnormality may precede functional abnormality in some patients, whereas the converse is true in other patients. Although the exact nature of the 'structure-function' relationship in primary open-angle glaucoma is still the topic of scientific debate and the subject of continuing research, this article aims to provide the clinician with an understanding of the past concepts and contemporary thinking in relation to the structure-function relationship in primary open-angle glaucoma.