Performance of Deep Learning Architectures and Transfer Learning for Detecting Glaucomatous Optic Neuropathy in Fundus Photographs

Christopher, Mark; Belghith, Akram; Bowd, Christopher; Proudfoot, James; Goldbaum, Michael H.; Weinreb, Robert N.; Girkin, Christopher A.; Liebmann, Jeffrey M.; Zangwill, Linda M.

doi:10.1038/s41598-018-35044-9

Cited by 245 publications

(212 citation statements)

References 49 publications

Supporting

Mentioning

203

Contrasting

Order By: Relevance

“…Some public data sets such as DRISHTI-GS provide glaucoma labels that were assigned based only on image characteristics. This has been also observed in private data sets such as those used by Christopher et al (2018) andLi et al (2018b), which were built using images from Internet that were manually graded based on their visual appeareance, without additional clinical information. Surprisingly, no information about the source of the diagnostic labels is provided in most of existing databases (see Table 1).…”

Section: Evaluation Protocolsmentioning

confidence: 70%

“…This is useful to prevent overfitting but limits the ability of the networks to learn rare, specific features. Alternatively, the studies by Christopher et al (2018), Li et al (2018a) and Orlando et al (2017b) used transfer learning methods, based on deeper architectures but pre-trained on non-medical data. Christopher et al (2018) fine-tuned a network initialized with weights learned from ImageNet to detect glaucomatous optic neuropathy.…”

Section: Automated Glaucoma Assessment: State-of-the-art and Current mentioning

confidence: 99%

“…Alternatively, the studies by Christopher et al (2018), Li et al (2018a) and Orlando et al (2017b) used transfer learning methods, based on deeper architectures but pre-trained on non-medical data. Christopher et al (2018) fine-tuned a network initialized with weights learned from ImageNet to detect glaucomatous optic neuropathy. Similarly, transfer learning was shown by Gómez-Valverde et al (2019) to outperform networks trained from scratch for glaucoma detection.…”

Section: Automated Glaucoma Assessment: State-of-the-art and Current mentioning

confidence: 99%

“…Most of the literature in glaucoma classification uses receiver-operating characteristic (ROC) curves (Davis and Goadrich, 2006) for evaluation, including the area under the curve (AUC) as a summary value (Chen et al, 2015a,b;Christopher et al, 2018;Fu et al, 2018;Gómez-Valverde et al, 2019;Orlando et al, 2017b;Li et al, 2018a,b;Liu et al, 2018;Pal et al, 2018). Sensitivity and specificity (Chen et al, 2015b;Christopher et al, 2018;Fu et al, 2018;Gómez-Valverde et al, 2019;Li et al, 2018a;Liu et al, 2018) are also used in different studies to complement the AUC when targetting binary classification outcomes. Accuracy was reported in (Cerentinia et al, 2018;Raghavendra et al, 2018) as another evaluation metric, although this metric might be biased if the proportion of non-glaucomatous images is significantly higher than the glaucomatous ones (Orlando et al, 2017a).…”

Section: Metricsmentioning

confidence: 99%

See 3 more Smart Citations

REFUGE Challenge: A unified framework for evaluating automated methods for glaucoma assessment from fundus photographs

Orlando

Breda

et al. 2020

Medical Image Analysis

526

248

View full text Add to dashboard Cite

Glaucoma is one of the leading causes of irreversible but preventable blindness in working age populations.Color fundus photography (CFP) is the most cost-effective imaging modality to screen for retinal disorders.However, its application to glaucoma has been limited to the computation of a few related biomarkers such as the vertical cup-to-disc ratio. Deep learning approaches, although widely applied for medical image analysis, have not been extensively used for glaucoma assessment due to the limited size of the available data sets. Furthermore, the lack of a standardize benchmark strategy makes difficult to compare existing methods in a uniform way. In order to overcome these issues we set up the Retinal Fundus Glaucoma Challenge, REFUGE (https://refuge.grand-challenge.org), held in conjunction with MICCAI 2018. The challenge consisted * Corresponding authors: Yanwu Xu (ywxu@ieee.org) and Xiulan Zhang (

show abstract

Section: Evaluation Protocolsmentioning

confidence: 70%

Section: Automated Glaucoma Assessment: State-of-the-art and Current mentioning

confidence: 99%

Section: Automated Glaucoma Assessment: State-of-the-art and Current mentioning

confidence: 99%

Section: Metricsmentioning

confidence: 99%

See 2 more Smart Citations

REFUGE Challenge: A unified framework for evaluating automated methods for glaucoma assessment from fundus photographs

Orlando

Breda

et al. 2020

Medical Image Analysis

526

248

View full text Add to dashboard Cite

show abstract

“…Li et al 13 applied a deep learning model using Inception-v3 23 architecture on a large dataset with about 40,000 fundus photographs and achieved an AUC of about 0.99 in detecting referable glauocma defined based on GON. In another study, Christopher et al 14 15 developed multiple deep learning models and used relatively small datasets with total of about 500 fundus photographs and achieved an accuracy of 92%. The reported accuracy of the previous deep learning models to diagnose glaucoma ranges from 0.83 to 0.98.…”

Section: Discussionmentioning

confidence: 99%

Predicting glaucoma prior to its onset using deep learning

Thakur

Goldbaum²,

Yousefi

2019

Preprint

View full text Add to dashboard Cite

AbstractPurposeTo assess the accuracy of deep learning models to predict glaucoma development from fundus photographs several years prior to disease onset.DesignA deep learning model for prediction of glaucomatous optic neuropathy or visual field abnormality from color fundus photographs.ParticipantsWe retrospectively included 66,721 fundus photographs from 3,272 eyes of 1,636 subjects to develop deep leaning models.MethodFundus photographs and visual fields were carefully examined by two independent readers from the optic disc and visual field reading centers of the ocular hypertension treatment study (OHTS). When an abnormality was detected by the readers, subject was recalled for re-testing to confirm the abnormality and further confirmation by an endpoint committee. Using OHTS data, deep learning models were trained and tested using 85% of the fundus photographs and further validated (re-tested) on the remaining (held-out) 15% of the fundus photographs.Main Outcome MeasuresAccuracy and area under the receiver-operating characteristic curve (AUC).ResultsThe AUC of the deep learning model in predicting glaucoma development 4-7 years prior to disease onset was 0.77 (95% confidence interval 0.75, 0.79). The accuracy of the model in predicting glaucoma development about 1-3 years prior to disease onset was 0.88 (0.86, 0.91). The accuracy of the model in detecting glaucoma after onset was 0.95 (0.94, 0.96).ConclusionsDeep learning models can predict glaucoma development prior to disease onset with reasonable accuracy. Eyes with visual field abnormality but not glaucomatous optic neuropathy had a higher tendency to be missed by deep learning algorithms.

show abstract

Development and Validation of a Deep Learning System to Detect Glaucomatous Optic Neuropathy Using Fundus Photographs

et al. 2019

Self Cite

View full text Add to dashboard Cite

IMPORTANCE A deep learning system (DLS) that could automatically detect glaucomatous optic neuropathy (GON) with high sensitivity and specificity could expedite screening for GON.OBJECTIVE To establish a DLS for detection of GON using retinal fundus images and glaucoma diagnosis with convoluted neural networks (GD-CNN) that has the ability to be generalized across populations. DESIGN, SETTING, AND PARTICIPANTSIn this cross-sectional study, a DLS for the classification of GON was developed for automated classification of GON using retinal fundus images obtained from the Chinese Glaucoma Study Alliance, the Handan Eye Study, and online databases. The researchers selected 241 032 images were selected as the training data set. The images were entered into the databases on June 9, 2009, obtained on July 11, 2018, and analyses were performed on December 15, 2018. The generalization of the DLS was tested in several validation data sets, which allowed assessment of the DLS in a clinical setting without exclusions, testing against variable image quality based on fundus photographs obtained from websites, evaluation in a population-based study that reflects a natural distribution of patients with glaucoma within the cohort and an additive data set that has a diverse ethnic distribution. An online learning system was established to transfer the trained and validated DLS to generalize the results with fundus images from new sources. To better understand the DLS decision-making process, a prediction visualization test was performed that identified regions of the fundus images utilized by the DLS for diagnosis. EXPOSURES Use of a deep learning system. MAIN OUTCOMES AND MEASURES Area under the receiver operating characteristics curve (AUC), sensitivity and specificity for DLS with reference to professional graders. RESULTS From a total of 274 413 fundus images initially obtained from CGSA, 269 601 images passed initial image quality review and were graded for GON. A total of 241 032 images (definite GON 29 865 [12.4%], probable GON 11 046 [4.6%], unlikely GON 200 121 [83%]) from 68 013 patients were selected using random sampling to train the GD-CNN model. Validation and evaluation of the GD-CNN model was assessed using the remaining 28 569 images from CGSA. The AUC of the GD-CNN model in primary local validation data sets was 0.996 (95% CI, 0.995-0.998), with sensitivity of 96.2% and specificity of 97.7%. The most common reason for both false-negative and false-positive grading by GD-CNN (51 of 119 [46.3%] and 191 of 588 [32.3%]) and manual grading (50 of 113 [44.2%] and 183 of 538 [34.0%]) was pathologic or high myopia.CONCLUSIONS AND RELEVANCE Application of GD-CNN to fundus images from different settings and varying image quality demonstrated a high sensitivity, specificity, and generalizability for detecting GON. These findings suggest that automated DLS could enhance current screening programs in a cost-effective and time-efficient manner.

show abstract

Performance of Deep Learning Architectures and Transfer Learning for Detecting Glaucomatous Optic Neuropathy in Fundus Photographs

Cited by 245 publications

References 49 publications

REFUGE Challenge: A unified framework for evaluating automated methods for glaucoma assessment from fundus photographs

REFUGE Challenge: A unified framework for evaluating automated methods for glaucoma assessment from fundus photographs

Predicting glaucoma prior to its onset using deep learning

Development and Validation of a Deep Learning System to Detect Glaucomatous Optic Neuropathy Using Fundus Photographs

Contact Info

Product

Resources

About