Dataset Information

Accuracy of deep learning, a machine learning technology, using ultra-wide-field fundus ophthalmoscopy for detecting idiopathic macular holes.

ABSTRACT: We aimed to investigate the detection of idiopathic macular holes (MHs) using ultra-wide-field fundus images (Optos) with deep learning, which is a machine learning technology. The study included 910 Optos color images (715 normal images, 195 MH images). Of these 910 images, 637 were learning images (501 normal images, 136 MH images) and 273 were test images (214 normal images and 59 MH images). We conducted training with a deep convolutional neural network (CNN) using the images and constructed a deep-learning model. The CNN exhibited high sensitivity of 100% (95% confidence interval CI [93.5-100%]) and high specificity of 99.5% (95% CI [97.1-99.9%]). The area under the curve was 0.9993 (95% CI [0.9993-0.9994]). Our findings suggest that MHs could be diagnosed using an approach involving wide angle camera images and deep learning.

SUBMITTER: Nagasawa T

PROVIDER: S-EPMC6201738 | biostudies-literature | 2018

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Accuracy of deep learning, a machine learning technology, using ultra-wide-field fundus ophthalmoscopy for detecting idiopathic macular holes.

Nagasawa Toshihiko T Tabuchi Hitoshi H Masumoto Hiroki H Enno Hiroki H Niki Masanori M Ohsugi Hideharu H Mitamura Yoshinori Y

PeerJ 20181022

We aimed to investigate the detection of idiopathic macular holes (MHs) using ultra-wide-field fundus images (Optos) with deep learning, which is a machine learning technology. The study included 910 Optos color images (715 normal images, 195 MH images). Of these 910 images, 637 were learning images (501 normal images, 136 MH images) and 273 were test images (214 normal images and 59 MH images). We conducted training with a deep convolutional neural network (CNN) using the images and constructed ...[more]

PMID: 30370184

Similar Datasets

Project description:BackgroundLamellar macular holes (LMHs) are small, partial-thickness defects of the macula defined by characteristic features on optical coherence tomography (OCT), including a newly recognised type of epiretinal membrane termed 'epiretinal proliferation'. There may be a rationale to recommend surgery for individuals with LMHs, particularly those with functional or anatomical deterioration, or poor baseline vision causing significant disability, to stabilise the LMH and prevent further visual deterioration; however, there is currently no evidence-based consensus.ObjectivesTo assess the effect of surgical interventions on post-operative visual and anatomical outcomes in people with a confirmed LMH.Search methodsWe searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE Ovid, Embase Ovid, Scopus SciVerse, ISRCTN registry, US National Institutes of Health Ongoing Trials Register, ClinicalTrials.gov and the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP). We also searched reference lists of included trials to identify other eligible trials which our search strategy may have missed. The date of the search was 20 July 2021.Selection criteriaWe included randomised controlled trials (RCTs) involving participants with a confirmed LMH diagnosis which reported one or more surgical intervention(s), alone or in combination, in at least one arm of the RCT.Data collection and analysisWe used standard methods as expected by Cochrane. Two study authors independently extracted data and assessed the risk of bias for included trials. Trial authors were contacted for further information and clarification.Main resultsA single RCT was eligible for inclusion. Thirty-six participants were randomised in a 2:1 ratio; 24 were allocated to undergo surgery (pars plana vitrectomy, peeling of the epiretial proliferation followed by fovea-sparing removal of the internal limiting membrane) and 12 (10 following two participant dropouts) to observation. Overall, the certainty of the evidence was low for all outcomes due to selection and detection bias, and the low number of participants enrolled in the study which may affect the accuracy of results and reliability of conclusions. At six-month follow-up, change in vision was better in the surgery group (-0.27 logMAR improvement) than observation (0.02 worsening) (mean difference (MD): -0.29 logMAR, 95% confidence intervals (CI): -0.33 to -0.25). Central retinal thickness increased in the surgery group over 6 months 126 μm increase) compared with observation group (decrease by 11μm) (MD: 137 μm, 95% CI: 125.87 μm to 148.13 μm). Finally, at six-month follow-up, retinal sensitivity was better in the surgery group (3.03 dB increase) compared with the observation group (0.06 dB decrease) (MD: 3.09 dB, 95% CI: 2.07 to 4.11 dB). Vision-related quality of life and metamorphopsia were not reported. No adverse outcomes or complications were reported in the study, however, authors could not provide information on whether any individuals developed deterioration in vision of 0.2 logMAR or worse.Authors' conclusionsThe included single trial demonstrated improvements in visual and anatomical outcome measures for participants with a LMH who underwent surgery compared with observation only. Therefore, we can conclude that participants who undergo surgery may achieve superior post-operative best corrected visual acuity and anatomical outcomes compared with observation only. However, the results of a single and small RCT provides limited evidence to support or refute surgery as an effective management option for LMHs. Future RCTs with a larger number of participants and with fewer methodological limitations and biases are necessary to inform future clinical practice.

Dataset Information

Accuracy of deep learning, a machine learning technology, using ultra-wide-field fundus ophthalmoscopy for detecting idiopathic macular holes.

Publications

Accuracy of deep learning, a machine learning technology, using ultra-wide-field fundus ophthalmoscopy for detecting idiopathic macular holes.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets