Dataset Information

Cross-sectional study: Does combining optical coherence tomography measurements using the 'Random Forest' decision tree classifier improve the prediction of the presence of perimetric deterioration in glaucoma suspects?

ABSTRACT:

Objectives

To develop a classifier to predict the presence of visual field (VF) deterioration in glaucoma suspects based on optical coherence tomography (OCT) measurements using the machine learning method known as the 'Random Forest' algorithm.

Design

Case-control study.

Participants

293 eyes of 179 participants with open angle glaucoma (OAG) or suspected OAG.

Interventions

Spectral domain OCT (Topcon 3D OCT-2000) and perimetry (Humphrey Field Analyser, 24-2 or 30-2 SITA standard) measurements were conducted in all of the participants. VF damage (Ocular Hypertension Treatment Study criteria (2002)) was used as a 'gold-standard' to classify glaucomatous eyes. The 'Random Forest' method was then used to analyse the relationship between the presence/absence of glaucomatous VF damage and the following variables: age, gender, right or left eye, axial length plus 237 different OCT measurements.

Main outcome measures

The area under the receiver operating characteristic curve (AROC) was then derived using the probability of glaucoma as suggested by the proportion of votes in the Random Forest classifier. For comparison, five AROCs were derived based on: (1) macular retinal nerve fibre layer (m-RNFL) alone; (2) circumpapillary (cp-RNFL) alone; (3) ganglion cell layer and inner plexiform layer (GCL+IPL) alone; (4) rim area alone and (5) a decision tree method using the same variables as the Random Forest algorithm.

Results

The AROC from the combined Random Forest classifier (0.90) was significantly larger than the AROCs based on individual measurements of m-RNFL (0.86), cp-RNFL (0.77), GCL+IPL (0.80), rim area (0.78) and the decision tree method (0.75; p<0.05).

Conclusions

Evaluating OCT measurements using the Random Forest method provides an accurate prediction of the presence of perimetric deterioration in glaucoma suspects.

SUBMITTER: Sugimoto K

PROVIDER: S-EPMC3796272 | biostudies-literature | 2013 Oct

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Cross-sectional study: Does combining optical coherence tomography measurements using the 'Random Forest' decision tree classifier improve the prediction of the presence of perimetric deterioration in glaucoma suspects?

Sugimoto Koichiro K Murata Hiroshi H Hirasawa Hiroyo H Aihara Makoto M Mayama Chihiro C Asaoka Ryo R

BMJ open 20131007 10

<h4>Objectives</h4>To develop a classifier to predict the presence of visual field (VF) deterioration in glaucoma suspects based on optical coherence tomography (OCT) measurements using the machine learning method known as the 'Random Forest' algorithm.<h4>Design</h4>Case-control study.<h4>Participants</h4>293 eyes of 179 participants with open angle glaucoma (OAG) or suspected OAG.<h4>Interventions</h4>Spectral domain OCT (Topcon 3D OCT-2000) and perimetry (Humphrey Field Analyser, 24-2 or 30-2 ...[more]

PMID: 24103806

Similar Datasets

Project description:Background: This study aimed to assess the differences in the average and sectoral peripapillary retinal nerve fiber layer (pRNFL), macular ganglion cell plus inner plexiform layer (mGCIPL), and macular ganglion cell complex (mGCC) thickness using optical coherence tomography (OCT) in patients with pre-perimetric glaucoma (PPG) compared to those with early perimetric glaucoma (EG) and ocular hypertension (OHT). Methods: A comprehensive literature search of the PubMed database, the Cochrane Library, and Embase was performed from inception to March 2021. The weighted mean difference (WMD) with the 95% confidence interval (CI) was pooled for continuous outcomes. Results: Twenty-three cross-sectional studies comprising 2,574 eyes (1,101 PPG eyes, 1,233 EG eyes, and 240 OHT eyes) were included in the systematic review and meta-analysis. The pooled results demonstrated that the average pRNFL (WMD = 8.22, 95% CI = 6.32-10.12, P < 0.00001), mGCIPL (WMD = 4.83, 95% CI = 3.43-6.23, P < 0.00001), and mGCC (WMD = 7.19, 95% CI = 4.52-9.85, P < 0.00001) were significantly thinner in patients with EG than in those with PPG. The sectoral thickness of pRNFL, mGCIPL, and mGCC were also significantly lower in the EG eyes. In addition, the average pRNFL and mGCC were significantly thinner in the PPG eyes than those in the OHT eyes (pRNFL: WMD = -8.57, 95% CI = -9.88 to -7.27, P < 0.00001; mGCC: WMD = -3.23, 95% CI = -6.03 to -0.44, P = 0.02). Similarly, the sectoral pRNFL and mGCC were also significantly thinner in the PPG eyes than those in the OHT eyes. Conclusion: OCT-based measurements of peripapillary and macular structural alterations can be used to distinguish PPG from EG and OHT, which can help understand the pathophysiology of glaucoma at earlier stages. Studies that employ clock hour classification methods and longitudinal studies are needed to verify our findings. Systematic Review Registration: https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=239798 CRD42021239798.

Project description:PurposeCompare the ability of peripapillary and macular structural parameters, vascular parameters, and their integration to discriminate among glaucoma, suspected glaucoma (GS), and healthy controls (HCs).MethodsIn this study, 196 eyes of 119 patients with glaucoma (n = 81), patients with GS (n = 48), and HCs (n = 67) underwent optical coherence tomography (OCT) and OCT angiography to measure peripapillary retinal nerve fiber layer (pRNFL), macular ganglion cell-inner plexiform layer (mGCIPL) thicknesses, radial peripapillary capillary perfusion density (RPC-PD), and macular GCIPL perfusion density (GCIPL-PD). Parameters were integrated regionally with logistic regression and globally with machine learning algorithms. Diagnostic performances were evaluated with area under the receiver operating characteristic (AUROC) curves.ResultsPatients with glaucoma had mild to moderate damage (median, -3.3 dB; interquartile range, -6.5 to -1.4). In discriminating between patients with glaucoma and the HCs, pRNFL thickness had higher AUROC curve values than RPC-PD for average (0.87 vs. 0.62; P < 0.001), superior (0.86 vs. 0.54; P < 0.001), inferior (0.90 vs. 0.71; P < 0.001), and temporal (0.65 vs. 0.51; P = 0.02) quadrants. mGCIPL thickness had higher AUROC curve values than GCIPL-PD for average (0.84 vs. 0.68; P < 0.001), superotemporal (0.76 vs. 0.65; P = 0.016), superior (0.72 vs. 0.57; P = 0.004), superonasal (0.70 vs. 0.56; P = 0.01), inferotemporal (0.90 vs. 0.72; P < 0.001), inferior (0.87 vs. 0.69; P < 0.001), and inferonasal (0.78 vs. 0.65, P = 0.012) sectors. All structural multisector indices had higher diagnostic ability than vascular ones (P < 0.001). Combined structural-vascular indices did not outperform structural indices. Similar results were found to discriminate glaucoma from GS.ConclusionsCombining structural and vascular parameters in a structural-vascular index does not improve diagnostic ability over structural parameters alone.Translational relevanceOCT angiography does not add additional benefit to structural OCT in early to moderate glaucoma diagnosis.

Project description:PurposeTo evaluate optic nerve vascular density using swept source optical coherence tomography angiography (OCTA) in patients with early primary open angle glaucoma (POAG), pre-perimetric glaucoma and normal eyes.MethodsThis is a prospective, observational study including 56 eyes in total and divided into 3 groups; 20 eyes with mild POAG, 20 pre-perimetric glaucoma eyes, and 16 age-matched normal eyes as controls. The optic disc region was imaged by a 1050-nm-wavelength swept-source OCT system (DRI OCT Triton, TOPCON). Vessel density was assessed as the ratio of the area occupied by the vessels in 3 distinct regions: 1) within the optic nerve head; 2) in the 3 mm papillary region around the optic disc; and 3) in the peripapillary region, defined as a 700-μm-wide elliptical annulus around the disc. The potential associations between vessel density and structural, functional measures were analyzed.ResultsThere was a statistically significant difference for the peripapillary vessel density, optic nerve head vessel density, and papillary vessel density among all the groups (p<0.001). Control eyes showed a significant difference for all measured vessel densities compared to glaucomatous eyes (p values from 0.001 to 0.024). There was a statistically significant difference between control and pre-perimetric glaucoma eyes for peripapillary, optic nerve head and papillary vessel density values (p values from 0.001 to 0.007). The optic nerve head vessel density, superior and inferior papillary area vessel density (Pearson r = 0.512, 0.436, 0.523 respectively) were highly correlated with mean overall, superior and inferior RNFL thickness in POAG eyes (p = 0.04, p = 0.02 and p = 0.04 respectively). Multiple linear regression analysis of POAG group showed that optic nerve head vessel density in POAG group was more strongly linked to RNFL thickness than to any other variables.ConclusionsEyes with mild POAG could be differentiated from pre-perimetric glaucoma eyes, which also could be differentiated from normal eyes using OCTA-derived retinal vessel density measurements.

Project description:Importance:Whether optical coherence tomography angiography (OCT-A) outperforms OCT to detect glaucoma remains inconclusive. Objective:To compare (1) the diagnostic performance for detection of glaucoma and (2) the structure-function association between inner macular vessel density and inner macular thickness. Design, Setting, and Participants:This cross-sectional study included 115 patients with glaucoma and 35 healthy individuals for measurements of retinal thickness and retinal vessel density, segmented between the anterior boundary of internal limiting membrane and the posterior boundary of the inner plexiform layer, over the 3 × 3-mm2 macula using swept-source OCT. All participants were Chinese. Visual sensitivity corresponding to the 3 × 3-mm2 macular region was expressed in unlogged 1/lambert for investigation of the structure-function associations. Diagnostic performance was evaluated with area under the receiver operating characteristic curves (AUCs). The study was conducted between January 12, 2016, and December 12, 2016. Main Outcomes and Measures:Area under the receiver operating characteristic curve and R2 analysis. Results:Of the 115 patients with glaucoma, 42 (36.5%) were women (mean [SD] age, 53.5 [13.4] years); of the 35 individuals with healthy eyes, 25 (71.4%) were women (age, 60.6 [5.9] years). Inner macular vessel density and thickness were 4.3% (95% CI, 2.4%-6.1%) and 21.1 μm (95% CI, 17.4-24.9 μm) smaller, respectively, in eyes with glaucoma compared with healthy eyes. The AUC of mean inner macular thickness for glaucoma detection was greater than that of mean inner macular vessel density (difference, 0.17; 95% CI, 0.01-0.31; P = .03). At 90% specificity, the sensitivity of mean inner macular thicknesses for detection of glaucoma was greater than that of mean inner macular vessel densities (difference, 29.2%; 95% CI, 11.5%-64.6%; P = .02). The strength of the structure-function association was stronger for mean inner macular thickness than mean inner macular vessel density in the linear (difference in R2 = 0.38; 95% CI, 0.22-0.54; P < .001) and nonlinear (difference in R2 = 0.36; 95% CI, 0.21-0.51; P < .001) regression models. Conclusions and Relevance:In this study, OCT measurement of inner macular thickness shows a higher diagnostic performance to detect glaucoma and a stronger structure-function association than the currently used OCT-A measurement of inner macular vessel density. These findings may suggest that OCT-A of the macula has a limited role in the diagnostic evaluation of glaucoma.

Dataset Information

Cross-sectional study: Does combining optical coherence tomography measurements using the 'Random Forest' decision tree classifier improve the prediction of the presence of perimetric deterioration in glaucoma suspects?

Objectives

Design

Participants

Interventions

Main outcome measures

Results

Conclusions

Publications

Cross-sectional study: Does combining optical coherence tomography measurements using the 'Random Forest' decision tree classifier improve the prediction of the presence of perimetric deterioration in glaucoma suspects?

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets