An Artificial Intelligence Approach to Detect Visual Field Progression in Glaucoma Based on Spatial Pattern Analysis.
Ontology highlight
ABSTRACT:
SUBMITTER: Wang M
PROVIDER: S-EPMC6348996 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:PurposeTo quantify the central visual field (VF) loss patterns in glaucoma using artificial intelligence.DesignRetrospective study.ParticipantsVFs of 8712 patients with 13 951 Humphrey 10-2 test results from 13 951 eyes for cross-sectional analyses, and 824 patients with at least 5 reliable 10-2 test results at 6-month intervals or more from 1191 eyes for longitudinal analyses.MethodsTotal deviation values were used to determine the central VF patterns using the most recent 10-2 test results. A 24-2 VF within a 3-month window of the 10-2 tests was used to stage eyes into mild, moderate, or severe functional loss using the Hodapp-Anderson-Parrish scale at baseline. Archetypal analysis was applied to determine the central VF patterns. Cross-validation was performed to determine the optimal number of patterns. Stepwise regression was applied to select the optimal feature combination of global indices, average baseline decomposition coefficients from central VFs archetypes, and other factors to predict central VF mean deviation (MD) slope based on the Bayesian information criterion (BIC).Main outcome measuresThe central VF patterns stratified by severity stage based on 24-2 test results and a model to predict the central VF MD change over time using baseline test results.ResultsFrom cross-sectional analysis, 17 distinct central VF patterns were determined for the 13 951 eyes across the spectrum of disease severity. These central VF patterns could be divided into isolated superior loss, isolated inferior loss, diffuse loss, and other loss patterns. Notably, 4 of the 5 patterns of diffuse VF loss preserved the less vulnerable inferotemporal zone, whereas they lost most of the remaining more vulnerable zone described by the Hood model. Inclusion of coefficients from central VF archetypical patterns strongly improved the prediction of central VF MD slope (BIC decrease, 35; BIC decrease of >6 indicating strong prediction improvement) than using only the global indices of 2 baseline VF results. Eyes with baseline VF results with more superonasal and inferonasal loss were more likely to show worsening MD over time.ConclusionsWe quantified central VF patterns in glaucoma, which were used to improve the prediction of central VF worsening compared with using only global indices.
| S-EPMC7246163 | biostudies-literature
Project description:ImportanceAlthough the central visual field (VF) in end-stage glaucoma may substantially vary among patients, structure-function studies and quality-of-life assessments are impeded by the lack of appropriate characterization of end-stage VF loss.ObjectiveTo provide a quantitative characterization and classification of central VF loss in end-stage glaucoma.Design, setting, and participantsThis retrospective cohort study collected data from 5 US glaucoma services from June 1, 1999, through October 1, 2014. A total of 2912 reliable 10-2 VFs of 1103 eyes from 1010 patients measured after end-stage 24-2 VFs with a mean deviation (MD) of -22 dB or less were included in the analysis. Data were analyzed from March 28, 2018, through May 23, 2019.Main outcomes and measuresCentral VF patterns were determined by an artificial intelligence algorithm termed archetypal analysis. Longitudinal analyses were performed to investigate whether the development of central VF defect mostly affects specific vulnerability zones.ResultsAmong the 1103 patients with the most recent VFs, mean (SD) age was 70.4 (14.3) years; mean (SD) 10-2 MD, -21.5 (5.6) dB. Fourteen central VF patterns were determined, including the most common temporal sparing patterns (304 [27.5%]), followed by mostly nasal loss (280 [25.4%]), hemifield loss (169 [15.3%]), central island (120 [10.9%]), total loss (91 [8.3%]), nearly intact field (56 [5.1%]), inferonasal quadrant sparing (42 [3.8%]), and nearly total loss (41 [3.7%]). Location-specific median total deviation analyses partitioned the central VF into a more vulnerable superonasal zone and a less vulnerable inferotemporal zone. At 1-year and 2-year follow-up, new defects mostly occurred in the more vulnerable zone. Initial encroachments on an intact central VF at follow-up were more likely to be from nasal loss (11 [18.4%]; P < .001). One of the nasal loss patterns had a substantial chance at 2-year follow-up (8 [11.0%]; P = .004) to shift to total loss, whereas others did not.Conclusions and relevanceIn this study, central VF loss in end-stage glaucoma was found to exhibit characteristic patterns that might be associated with different subtypes. Initial central VF loss is likely to be nasal loss, and 1 specific type of nasal loss is likely to develop into total loss.
| S-EPMC6990977 | biostudies-literature
Project description:PurposeTo investigate clinical factors associated with foveal avascular zone (FAZ) parameters obtained using OCT angiography (OCTA) with assistance from a previously developed artificial intelligence (AI) platform in eyes with open-angle glaucoma (OAG).DesignRetrospective longitudinal.ParticipantsThis study followed up 885 eyes of 558 patients with OAG for ≥ 2 years; all eyes underwent ≥ 5 Humphrey visual-field (VF) tests and had 3.0 × 3.0 mm macular OCTA scans available.MethodsAverage total deviation (TD) in the superior, superocentral, inferocentral, and inferior sectors of the Humphrey 24-2 program was calculated. We collected 3.0 × 3.0 mm macular OCTA images from each patient and used a previously developed AI platform with these images to obtain FAZ parameters, including FAZ area, FAZ circularity index (CI), and FAZ perimeter. Multivariable linear mixed-effects models were used to analyze the relationship between FAZ parameters, TD or TD slope in each quadrant, and systemic factors, adjusting for potential confounding factors, including axial length.Main outcome measuresOphthalmic and systemic variables, FAZ parameters, and TD or TD slope in each quadrant.ResultsThe multivariable model showed that FAZ parameters were correlated with both TD and TD slope in the inferocentral quadrant (β = -0.244 - 0.168, P < 0.001). Both upper-half and lower-half FAZ parameters were better associated with TD-inferocentral and TD-inferocentral slope than TD-superocentral or TD-superocentral slope in terms of β size and statistical significance, indicating that there was no evident vertical anatomical correspondence between TD in the central quadrant and FAZ parameters. Foveal avascular zone area enlargement was associated with female gender (β = 0.242, P = 0.003). Loss of FAZ circularity was associated with both aging and comorbid sleep apnea syndrome (SAS) (yes: 1, no: 0) (β = -0.188, P < 0.001; β = -0.261, P = 0.031, respectively). Foveal avascular zone perimeter elongation was associated with aging and female gender (β = 0.084, P = 0.040; β = 0.168, P = 0.042, respectively).ConclusionsArtificial intelligence-assisted OCTA-measured FAZ enlargement and irregular shape might be good markers of ocular hypoperfusion and associated inferocentral VF defect progression in eyes with OAG. Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article.
| S-EPMC10960060 | biostudies-literature
Project description:PurposePrior artificial intelligence (AI) models for predicting glaucoma progression have used traditional classifiers that do not consider the longitudinal nature of patients' follow-up. In this study, we developed survival-based AI models for predicting glaucoma patients' progression to surgery, comparing performance of regression-, tree-, and deep learning-based approaches.DesignRetrospective observational study.SubjectsPatients with glaucoma seen at a single academic center from 2008 to 2020 identified from electronic health records (EHRs).MethodsFrom the EHRs, we identified 361 baseline features, including demographics, eye examinations, diagnoses, and medications. We trained AI survival models to predict patients' progression to glaucoma surgery using the following: (1) a penalized Cox proportional hazards (CPH) model with principal component analysis (PCA); (2) random survival forests (RSFs); (3) gradient-boosting survival (GBS); and (4) a deep learning model (DeepSurv). The concordance index (C-index) and mean cumulative/dynamic area under the curve (mean AUC) were used to evaluate model performance on a held-out test set. Explainability was investigated using Shapley values for feature importance and visualization of model-predicted cumulative hazard curves for patients with different treatment trajectories.Main outcome measuresProgression to glaucoma surgery.ResultsOf the 4512 patients with glaucoma, 748 underwent glaucoma surgery, with a median follow-up of 1038 days. The DeepSurv model performed best overall (C-index, 0.775; mean AUC, 0.802) among the models studied in this article (CPH with PCA: C-index, 0.745; mean AUC, 0.780; RSF: C-index, 0.766; mean AUC, 0.804; GBS: C-index, 0.764; mean AUC, 0.791). Predicted cumulative hazard curves demonstrate how models could distinguish between patient who underwent early surgery and patients who underwent surgery after > 3000 days of follow-up or no surgery.ConclusionsArtificial intelligence survival models can predict progression to glaucoma surgery using structured data from EHRs. Tree-based and deep learning-based models performed better at predicting glaucoma progression to surgery than the CPH regression model, potentially because of their better suitability for high-dimensional data sets. Future work predicting ophthalmic outcomes should consider using tree-based and deep learning-based survival AI models. Additional research is needed to develop and evaluate more sophisticated deep learning survival models that can incorporate clinical notes or imaging.Financial disclosuresProprietary or commercial disclosure may be found after the references.
| S-EPMC10320266 | biostudies-literature
Project description:PurposeThe purpose of this study was to classify the spatial patterns of retinal nerve fiber layer thickness (RNFLT) and assess their associations with visual field (VF) loss in glaucoma.MethodsWe used paired reliable 24-2 VFs and optical coherence tomography scans of 691 eyes from 691 patients. The RNFLT maps were used to determine the RNFLT patterns (RPs) by non-negative matrix factorization (NMF). The RPs were correlated with mean deviation (MD), spherical equivalent (SE), and major blood vessel locations. The RPs were further used to predict the 52 total deviation (TD) values by linear regression compared with models using 24 15-degree sectors. Last, we associated the RPs with average TDs of the central upper two locations (C2-TD). Stepwise regression was applied to remove redundant features.ResultsNMF highlighted 16 distinct RPs. Twelve RPs had arcuate-like informative zones (iZones): six with superior iZones, five with inferior iZones, and one RP with a bi-hemifield iZone, and four with non-arcuate-like temporal or nasal iZones. Twelve, nine, nine, and nine RPs were significantly (P < 0.05) correlated to MD, SE, and superior and inferior artery locations, respectively. Using RPs significantly (P < 0.05) improved the prediction of 52 TDs compared with using 24 15-degree sectors. Using RPs significantly (P < 0.001) improved the C2-TD prediction related to thinning in the inferior vulnerability zone compared with using the 24 sectoral RNFLTs.ConclusionsUsing RPs improved the VF prediction compared with using sectoral RNFLTs.Translational relevanceThe RPs characterizing both pathological and anatomical variations can potentially assist clinicians better assess RNFLT loss.
| S-EPMC7453051 | biostudies-literature
Project description:To explore visual field (VF) progression in a cohort of secondary care-treated glaucoma and ocular hypertensive (OHT) patients.We extracted VFs from our database drawn from our normal clinical practice. VF series from 4177 eyes from 2208 patients who had five or more VFs were obtained, the 'better' eye was selected and the rate of VF progression was calculated using mean deviation (MD) data.The median rate of progression for the whole sample was -0.1 dB/year (interquartile range (IQR) -4 to 0 dB/year) over a median of 6.7 years (IQR 4.9-8.7). Of 2208 patients, 477 (21.2%) progressed at > -0.5 dB/year; 46 (2.1%) progressed at >-2.0 dB/year. Of those with a 'final MD' of worse than -10 dB (N=244) in their better eye; 14.0% were 'fast progressors' (>-2 dB/year), 33.7% 'moderate progressors' (-1 to -2 dB/year), and 28.8% 'slow progressors' (-0.3 dB to -1 dB/year). Of those with 'initial MD' better than -3 dB and those with worse than -3 dB, 31/1679 (1.8%) and 213/529 (40.3%) respectively, had a final MD of worse than -10 dB.Fast progressors, while important, are relatively rare. Moderate and slow progressors make up the majority of the progressing population within this data set. The risk of significant visual loss is much higher in those with initial damage. With increasing life expectancy, moderate and slow progressors may become increasingly clinically important.
| S-EPMC4135248 | biostudies-other
Project description:PurposeIt has been suggested that the detection of visual field progression can be improved by modeling statistical properties of the data such as the increasing retest variability and the spatial correlation among visual field locations. We compared a method that models those properties, Analysis with Non-Stationary Weibull Error Regression and Spatial Enhancement (ANSWERS), against a simpler one that does not, Permutation of Pointwise Linear Regression (PoPLR).MethodsVisual field series from three independent longitudinal studies in patients with glaucoma were used to compare the positive rate of PoPLR and ANSWERS. To estimate the false-positive rate, the same visual field series were randomly re-ordered in time. The first dataset consisted of series of 7 visual fields from 101 eyes, the second consisted of series of 9 visual fields from 150 eyes, and the third consisted of series of more than 9 visual fields (17.5 on average) from 139 eyes.ResultsFor a statistical significance of 0.05, the false-positive rates for ANSWERS were about 3 times greater than expected at 15%, 17%, and 16%, respectively, whereas for PoPLR they were 7%, 3%, and 6%. After equating the specificities at 0.05 for both models, positive rates for ANSWERS were 16%, 25%, and 38%, whereas for PoPLR they were 12%, 33%, and 49%, or about 5% greater on average (95% confidence interval = -1% to 11%).ConclusionsDespite being simpler and less computationally demanding, PoPLR was at least as sensitive to deterioration as ANSWERS once the specificities were equated.Translational relevanceClose control of false-positive rates is key when visual fields of patients are analyzed for change in both clinical practice and clinical trials.
| S-EPMC8684309 | biostudies-literature
Project description:PurposeTo evaluate the association of macular superficial vessel density (SVD) and projection-resolved deep vessel density (DVD) with past visual field (VF) progression in patients with primary open-angle glaucoma.DesignRetrospective cohort.MethodsIn this longitudinal study, 208 eyes of 147 patients with glaucoma from the Diagnostics Innovations in Glaucoma Study were included. Eligible participants were required to have at least five 24-2 VF tests over a minimum follow-up period of 3 years before macular optical coherence tomography angiography imaging. VF progression was defined based on both event-based pointwise linear regression and trend-based methods. The association of macular SVD and DVD with the probability and rate of past VF progression was evaluated using a linear mixed effects model.ResultsFifty-two (25%) eyes had VF progression based on the pointwise linear regression based criterion at the end of a mean ± standard deviation follow-up duration of 6.9 ± 1.2 years. In the event-based multivariable analysis, a lower baseline SVD was associated with a higher likelihood of past VF progression (odds ratio per 1% lower. 1.28; 95% confidence interval, 1.02-1.59). Similarly, in the trend-based multivariable analysis, lower macular SVD was associated with a faster past rate of mean deviation decline (coefficient = -0.03 dB/year; 95% confidence interval, -0.04 to -0.01). Event-based and trend-based analyses found no significant associations for macular DVD with the likelihood/rate of past VF progression (P > .05).ConclusionsLower macular SVD, and not DVD, was associated with a higher probability of past VF progression. Macular optical coherence tomography angiography imaging shows promise for identifying eyes at risk of VF progression in patients with glaucoma.
| S-EPMC9469932 | biostudies-literature
Project description:ImportanceRates of visual field (VF) progression vary among patients with glaucoma. Knowing the rate of progression of individual patients would allow appropriately aggressive therapy for patients with high rates of visual loss and protect those with low rates from unnecessary therapy.ObjectiveTo compare 3 pointwise methods of estimating the rate of VF progression in glaucoma.Design, setting, and participantsThis retrospective, observational cohort study included 729 eyes of 567 consecutive patients with primary open-angle glaucoma who had at least 6 reliable VFs and at least 3 years of follow-up. One hundred seventy-six patients (257 eyes) were treated at a tertiary glaucoma center; in addition, data were collected from 391 participants (472 eyes) in the Advanced Glaucoma Intervention Study. Data were collected from May 1988 to November 2004 and analyzed from October 2018 to February 2019.ExposuresEstimates of VF progression were measured with guided progression analysis (GPA), pointwise linear regression (PLR), and the glaucoma rate index (GRI). A subgroup analysis was performed in a subset of patients with likely VF progression and likely VF stability.Main outcomes and measuresProportion of VF series detected as progressing, estimates of false-positive proportions, time to detect progression, and agreement among measures.ResultsAmong the 567 patients included in the analysis, mean (SD) age was 65.6 (9.7) years, 300 (52.9%) were female, and 295 (52.0%) were white. The median baseline mean deviation was -6.7 (interquartile range [IQR], -11.6 to -3.5) dB; the median follow-up time, 8.9 (IQR, 7.3-10.4) years. The proportion of eyes labeled as progressing was 27.7% according to the GPA, 33.5% according to the PLR, and 52.9% according to the GRI; pairwise differences for GRI vs PLR were 20% (95% CI, 17%-23%); for GRI vs GPA, 25% (95% CI, 22%-29%); and for PLR vs GPA, 6% (95% CI, 3%-9%; P < .001 for all comparisons, McNemar test). The shortest median time to progression was with the GRI (8.8 [IQR, 2.4-10.5 years), compared with the GPA and PLR (both >16 years). The hazard ratio of VF progression for GRI vs PLR (reference) was 11.3 (95% CI, 9.2-13.7); for GRI vs GPA (reference), 18.1 (95% CI, 14.5-22.6); and for PLR vs GPA (reference), 1.5 (95% CI, 1.3-1.9; P < .001 for all comparisons, Cox proportional hazards regression). These results held in the subgroup with likely progression; the proportions of progressing eyes were 73.7% (115 of 156) for GPA, 81.4% (127 of 156) for PLR, and 92.9% (145 of 156) for GRI. Pairwise difference for GRI vs PLR was 11.5% (95% CI, 7.4%-17.6%; P < .001, McNemar test); for GRI vs GPA, 19.2% (95% CI, 12.6%-26.4%; P < .001, McNemar test); and for PLR vs GPA, 7.7% (95% CI, 0.3%-15.7%; P = .08, McNemar test).Conclusions and relevanceThese results suggest GRI can detect long-term VF progression in glaucoma earlier than PLR or GPA. Validation with prospective designs may strengthen the generalizability and value of this method.
| S-EPMC7118669 | biostudies-literature
Project description:PurposeThe purpose of this study was to develop a model to predict whether or not glaucoma will progress to the point of requiring surgery within the following year, using data from electronic health records (EHRs), including both structured data and free-text progress notes.MethodsA cohort of adult glaucoma patients was identified from the EHR at Stanford University between 2008 and 2020, with data including free-text clinical notes, demographics, diagnosis codes, prior surgeries, and clinical information, including intraocular pressure, visual acuity, and central corneal thickness. Words from patients' notes were mapped to ophthalmology domain-specific neural word embeddings. Word embeddings and structured clinical data were combined as inputs to deep learning models to predict whether a patient would undergo glaucoma surgery in the following 12 months using the previous 4-12 months of clinical data. We also evaluated models using only structured data inputs (regression-, tree-, and deep-learning-based models) and models using only text inputs.ResultsOf the 3,469 glaucoma patients included in our cohort, 26% underwent surgery. The baseline penalized logistic regression model achieved an area under the receiver operating curve (AUC) of 0.873 and F1 score of 0.750, compared with the best tree-based model (random forest, AUC 0.876; F1 0.746), the deep learning structured features model (AUC 0.885; F1 0.757), the deep learning clinical free-text features model (AUC 0.767; F1 0.536), and the deep learning model with both the structured clinical features and free-text features (AUC 0.899; F1 0.745).DiscussionFusion models combining text and EHR structured data successfully and accurately predicted glaucoma progression to surgery. Future research incorporating imaging data could further optimize this predictive approach and be translated into clinical decision support tools.
| S-EPMC10133544 | biostudies-literature