scholarly journals Classifying central serous chorioretinopathy subtypes with a deep neural network using optical coherence tomography images: a cross-sectional study

2022 ◽  
Vol 12 (1) ◽  
Author(s):  
Jeewoo Yoon ◽  
Jinyoung Han ◽  
Junseo Ko ◽  
Seong Choi ◽  
Ji In Park ◽  
...  
Keyword(s):  
Neural Network ◽  
Optical Coherence Tomography ◽  
Deep Learning ◽  
Central Serous Chorioretinopathy ◽  
Cross Validation ◽  
Learning Systems ◽  
Test Accuracy ◽  
Optical Coherence ◽  
Proposed Model ◽  
Sd Oct

AbstractCentral serous chorioretinopathy (CSC) is the fourth most common retinopathy and can reduce quality of life. CSC is assessed using optical coherence tomography (OCT), but deep learning systems have not been used to classify CSC subtypes. This study aimed to build a deep learning system model to distinguish CSC subtypes using a convolutional neural network (CNN). We enrolled 435 patients with CSC from a single tertiary center between January 2015 and January 2020. Data from spectral domain OCT (SD-OCT) images of the patients were analyzed using a deep CNN. Five-fold cross-validation was employed to evaluate the model’s ability to discriminate acute, non-resolving, inactive, and chronic atrophic CSC. We compared the performances of the proposed model, Resnet-50, Inception-V3, and eight ophthalmologists. Overall, 3209 SD-OCT images were included. The proposed model showed an average cross-validation accuracy of 70.0% (95% confidence interval [CI], 0.676–0.718) and the highest test accuracy was 73.5%. Additional evaluation in an independent set of 104 patients demonstrated the reliable performance of the proposed model (accuracy: 76.8%). Our model could classify CSC subtypes with high accuracy. Thus, automated deep learning systems could be useful in the classification and management of CSC.

scholarly journals Distinguishing retinal angiomatous proliferation from polypoidal choroidal vasculopathy with a deep neural network based on optical coherence tomography

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Daniel Duck-Jin Hwang ◽  
Seong Choi ◽  
Junseo Ko ◽  
Jeewoo Yoon ◽  
Ji In Park ◽  
...  
Keyword(s):  
Neural Network ◽  
Optical Coherence Tomography ◽  
Deep Learning ◽  
Polypoidal Choroidal Vasculopathy ◽  
Retinal Angiomatous Proliferation ◽  
Optical Coherence ◽  
Proposed Model ◽  
Tertiary Center ◽  
Choroidal Vasculopathy ◽  
Sd Oct

AbstractThis cross-sectional study aimed to build a deep learning model for detecting neovascular age-related macular degeneration (AMD) and to distinguish retinal angiomatous proliferation (RAP) from polypoidal choroidal vasculopathy (PCV) using a convolutional neural network (CNN). Patients from a single tertiary center were enrolled from January 2014 to January 2020. Spectral-domain optical coherence tomography (SD-OCT) images of patients with RAP or PCV and a control group were analyzed with a deep CNN. Sensitivity, specificity, accuracy, and area under the receiver operating characteristic curve (AUROC) were used to evaluate the model’s ability to distinguish RAP from PCV. The performances of the new model, the VGG-16, Resnet-50, Inception, and eight ophthalmologists were compared. A total of 3951 SD-OCT images from 314 participants (229 AMD, 85 normal controls) were analyzed. In distinguishing the PCV and RAP cases, the proposed model showed an accuracy, sensitivity, and specificity of 89.1%, 89.4%, and 88.8%, respectively, with an AUROC of 95.3% (95% CI 0.727–0.852). The proposed model showed better diagnostic performance than VGG-16, Resnet-50, and Inception-V3 and comparable performance with the eight ophthalmologists. The novel model performed well when distinguishing between PCV and RAP. Thus, automated deep learning systems may support ophthalmologists in distinguishing RAP from PCV.

scholarly journals Optical coherence tomography-based deep-learning model for detecting central serous chorioretinopathy

2020 ◽  
Vol 10 (1) ◽  
Author(s):  
Jeewoo Yoon ◽  
Jinyoung Han ◽  
Ji In Park ◽  
Joon Seo Hwang ◽  
Jeong Mo Han ◽  
...  
Keyword(s):  
Optical Coherence Tomography ◽  
Deep Learning ◽  
Sensitivity And Specificity ◽  
Central Serous Chorioretinopathy ◽  
Characteristic Curve ◽  
Posterior Pole ◽  
Learning System ◽  
Control Group ◽  
Optical Coherence ◽  
Sd Oct

Abstract Central serous chorioretinopathy (CSC) is a common condition characterized by serous detachment of the neurosensory retina at the posterior pole. We built a deep learning system model to diagnose CSC, and distinguish chronic from acute CSC using spectral domain optical coherence tomography (SD-OCT) images. Data from SD-OCT images of patients with CSC and a control group were analyzed with a convolutional neural network. Sensitivity, specificity, accuracy, and area under the receiver operating characteristic curve (AUROC) were used to evaluate the model. For CSC diagnosis, our model showed an accuracy, sensitivity, and specificity of 93.8%, 90.0%, and 99.1%, respectively; AUROC was 98.9% (95% CI, 0.983–0.995); and its diagnostic performance was comparable with VGG-16, Resnet-50, and the diagnoses of five different ophthalmologists. For distinguishing chronic from acute cases, the accuracy, sensitivity, and specificity were 97.6%, 100.0%, and 92.6%, respectively; AUROC was 99.4% (95% CI, 0.985–1.000); performance was better than VGG-16 and Resnet-50, and was as good as the ophthalmologists. Our model performed well when diagnosing CSC and yielded highly accurate results when distinguishing between acute and chronic cases. Thus, automated deep learning system algorithms could play a role independent of human experts in the diagnosis of CSC.

Discriminating glaucomatous and compressive optic neuropathy on spectral-domain optical coherence tomography with deep learning classifier

2020 ◽  
Vol 104 (12) ◽  
pp. 1717-1723 ◽  
Author(s):  
Jinho Lee ◽  
Jin-Soo Kim ◽  
Haeng Jin Lee ◽  
Seong-Joon Kim ◽  
Young Kook Kim ◽  
...  
Keyword(s):  
Optical Coherence Tomography ◽  
Deep Learning ◽  
Optic Neuropathy ◽  
Spectral Domain ◽  
Optical Coherence ◽  
Compressive Optic Neuropathy ◽  
Diagnostic Parameters ◽  
Learning Classifier ◽  
Oct Image ◽  
Sd Oct

Background/aimsTo assess the performance of a deep learning classifier for differentiation of glaucomatous optic neuropathy (GON) from compressive optic neuropathy (CON) based on ganglion cell–inner plexiform layer (GCIPL) and retinal nerve fibre layer (RNFL) spectral-domain optical coherence tomography (SD-OCT).MethodsEighty SD-OCT image sets from 80 eyes of 80 patients with GON along with 81 SD-OCT image sets from 54 eyes of 54 patients with CON were compiled for the study. The bottleneck features extracted from the GCIPL thickness map, GCIPL deviation map, RNFL thickness map and RNFL deviation map were used as predictors for the deep learning classifier. The area under the receiver operating characteristic curve (AUC) was calculated to validate the diagnostic performance. The AUC with the deep learning classifier was compared with those for conventional diagnostic parameters including temporal raphe sign, SD-OCT thickness profile and standard automated perimetry.ResultsThe deep learning system achieved an AUC of 0.990 (95% CI 0.982 to 0.999) with a sensitivity of 97.9% and a specificity of 92.6% in a fivefold cross-validation testing, which was significantly larger than the AUCs with the other parameters: 0.804 (95% CI 0.737 to 0.872) with temporal raphe sign, 0.815 (95% CI 0.734 to 0.896) with superonasal GCIPL and 0.776 (95% CI 0.691 to 0.860) with superior GCIPL thicknesses (all p<0.001).ConclusionThe deep learning classifier can outperform the conventional diagnostic parameters for discrimination of GON and CON on SD-OCT.

Deep learning model to predict visual field in central 10° from optical coherence tomography measurement in glaucoma

2020 ◽  
pp. bjophthalmol-2019-315600
Author(s):  
Yohei Hashimoto ◽  
Ryo Asaoka ◽  
Taichi Kiwaki ◽  
Hiroki Sugiura ◽  
Shotaro Asano ◽  
...  
Keyword(s):  
Optical Coherence Tomography ◽  
Deep Learning ◽  
Visual Field ◽  
Retinal Pigment ◽  
Prediction Performance ◽  
Support Vector ◽  
Inner Plexiform Layer ◽  
Optical Coherence ◽  
Data Set ◽  
Sd Oct

Background/AimTo train and validate the prediction performance of the deep learning (DL) model to predict visual field (VF) in central 10° from spectral domain optical coherence tomography (SD-OCT).MethodsThis multicentre, cross-sectional study included paired Humphrey field analyser (HFA) 10-2 VF and SD-OCT measurements from 591 eyes of 347 patients with open-angle glaucoma (OAG) or normal subjects for the training data set. We trained a convolutional neural network (CNN) for predicting VF threshold (TH) sensitivity values from the thickness of the three macular layers: retinal nerve fibre layer, ganglion cell layer+inner plexiform layer and outer segment+retinal pigment epithelium. We implemented pattern-based regularisation on top of CNN to avoid overfitting. Using an external testing data set of 160 eyes of 131 patients with OAG, the prediction performance (absolute error (AE) and R2 between predicted and actual TH values) was calculated for (1) mean TH in whole VF and (2) each TH of 68 points. For comparison, we trained support vector machine (SVM) and multiple linear regression (MLR).ResultsAE of whole VF with CNN was 2.84±2.98 (mean±SD) dB, significantly smaller than those with SVM (5.65±5.12 dB) and MLR (6.96±5.38 dB) (all, p<0.001). Mean of point-wise mean AE with CNN was 5.47±3.05 dB, significantly smaller than those with SVM (7.96±4.63 dB) and MLR (11.71±4.15 dB) (all, p<0.001). R2 with CNN was 0.74 for the mean TH of whole VF, and 0.44±0.24 for the overall 68 points.ConclusionDL model showed considerably accurate prediction of HFA 10-2 VF from SD-OCT.

scholarly journals Deep-Learning Based, Automated Segmentation of Macular Edema in Optical Coherence Tomography

2017 ◽  
Author(s):  
Cecilia S. Lee ◽  
Ariel J. Tyring ◽  
Nicolaas P. Deruyter ◽  
Yue Wu ◽  
Ariel Rokem ◽  
...  
Keyword(s):  
Neural Network ◽  
Computer Vision ◽  
Optical Coherence Tomography ◽  
Deep Learning ◽  
Automated Segmentation ◽  
Optical Coherence ◽  
Dice Coefficient ◽  
Clinical Images ◽  
Intraretinal Fluid ◽  
Segmented Images

AbstractEvaluation of clinical images is essential for diagnosis in many specialties and the development of computer vision algorithms to analyze biomedical images will be important. In ophthalmology, optical coherence tomography (OCT) is critical for managing retinal conditions. We developed a convolutional neural network (CNN) that detects intraretinal fluid (IRF) on OCT in a manner indistinguishable from clinicians. Using 1,289 OCT images, the CNN segmented images with a 0.911 cross-validated Dice coefficient, compared with segmentations by experts. Additionally, the agreement between experts and between experts and CNN were similar. Our results reveal that CNN can be trained to perform automated segmentations.

scholarly journals Fundus Autofluorescence and Optical Coherence Tomography Characteristics in Different Stages of Central Serous Chorioretinopathy

2021 ◽  
Vol 2021 ◽  
pp. 1-9
Author(s):  
Mary Ho ◽  
Stephanie H. W. Kwok ◽  
Andrew C. Y. Mak ◽  
Frank H. P. Lai ◽  
Danny S. C. Ng ◽  
...  
Keyword(s):  
Optical Coherence Tomography ◽  
Central Serous Chorioretinopathy ◽  
Morphological Changes ◽  
Fundus Autofluorescence ◽  
Serous Retinal Detachment ◽  
Dependent Manner ◽  
Optical Coherence ◽  
Photoreceptor Layer ◽  
Oct Imaging ◽  
Sd Oct

Objective. To describe the morphological changes on fundus autofluorescence (FAF) and spectral-domain optical coherence tomography (SD-OCT) imaging at different chronicity of central serous chorioretinopathy (CSC). Methods. This cross-sectional study included patients with CSC of different chronicity. Changes in FAF scans and morphological changes on SD-OCT were evaluated and compared at different stages of CSC. Results. Sixty-nine patients were enrolled in the study, with a mean age of 52.1 ± 11.8 years. A distinct hypoautofluorescence (AF) pattern was observed at the leakage point in acute CSC (100%). The leakage site was indistinguishable in 48% of the patients with late-chronic CSC. The majority of acute CSC patients showed hyper-AF in the area of serous retinal detachment (SRD), which persisted in the early-chronic stage of CSC. In late-chronic CSC, many cases of hypo-AF (22.2%) and mixed-pattern AF (14.8%) were observed. SD-OCT revealed evolving features of retinal pigment epithelium (RPE) abnormalities in a time-dependent manner: from peaked PEDs in acute CSC to low-lying PEDs in early-chronic CSC and, eventually, flat, irregular PEDs in late-chronic CSC. The average thickness of the photoreceptor layer (inner and outer segment; IS/OS) was 79 μm in the acute group and 55.2 μm in the chronic group. The photoreceptor layer (IS/OS) height was positively associated with visual acuity ( p = 0.002 ). Conclusion. Different stages of CSC present different patterns on FAF and SD-OCT imaging. Chronicity of CSC can be estimated using specific features in these images. Photoreceptor layer (IS/OS) height acts as a good and objective predictor of visual outcomes in CSC patients.

743Combining deep learning method with optical coherence tomography for ablation lesion assessment

EP Europace ◽  
2020 ◽  
Vol 22 (Supplement_1) ◽  
Author(s):  
D Liang ◽  
A Haeberlin
Keyword(s):  
Neural Network ◽  
Optical Coherence Tomography ◽  
Deep Learning ◽  
Convolutional Neural Network ◽  
Medical Image Analysis ◽  
Ground Truth ◽  
Optical Coherence ◽  
Learning Method ◽  
Ablation Lesion ◽  
Learning Methods

Abstract Background The immediate effect of radiofrequency catheter ablation (RFA) on the tissue is not directly visualized. Optical coherence tomography (OCT) is an imaging technique that uses light to capture histology-like images with a penetration depth of 1-3 mm in the cardiac tissue. There are two specific features of ablation lesions in the OCT images: the disappearance of birefringence artifacts in the lateral and sudden decrease of signal at the bottom (Figure panel A and D). These features can not only be used to recognize the ablation lesions from the OCT images by eye, but also be used to train a machine learning model for automatic lesion segmentation. In recent years, deep learning methods, e.g. convolutional neural networks, have been used in medical image analysis and greatly increased the accuracy of image segmentation. We hypothesize that using a convolutional neural network, e.g. U-Net, can locate and segment the ablation lesions in the OCT images. Purpose To investigate whether a deep learning method such as a convolutional neural network optimized for biomedical image processing, could be used to segment ablation lesions in OCT images automatically. Method 8 OCT datasets with ablation lesions were used for training the convolutional neural network (U-Net model). After training, the model was validated by two new OCT datasets. Dice coefficients were calculated to evaluate spatial overlap between the predictions and the ground truth segmentations, which were manually segmented by the researchers (its value ranges from 0 to 1, and "1" means perfect segmentation). Results The U-Net model could predict the central parts of lesions automatically and accurately (Dice coefficients are 0.933 and 0.934), compared with the ground truth segmentations (Figure panel B and E). These predictions could reveal the depths and diameters of the ablation lesions correctly (Figure panel C and F). Conclusions  Our results showed that deep learning could facilitate ablation lesion identification and segmentation in OCT images. Deep learning methods, integrated in an OCT system, might enable automatic and precise ablation lesion visualization, which may help to assess ablation lesions during radiofrequency ablation procedures with great precision. Figure legend Panel A and D: the central OCT images of the ablation lesions. The blue arrows indicate the lesion bottom, where the image intensity suddenly decreases. The white arrows indicate the birefringence artifacts (the black bands in the grey regions). Panel B and E: the ground true segmentations of lesions in panel A and D. Panel C and F: the predictions by U-Net model of the lesions in panel A and D. A scale bar representing 500 μm is shown in each panel. Abstract Figure

Dual-input convolutional neural network for glaucoma diagnosis using spectral-domain optical coherence tomography

2020 ◽  
pp. bjophthalmol-2020-316274
Author(s):  
Sukkyu Sun ◽  
Ahnul Ha ◽  
Young Kook Kim ◽  
Byeong Wook Yoo ◽  
Hee Chan Kim ◽  
...  
Keyword(s):  
Neural Network ◽  
Optical Coherence Tomography ◽  
Deep Learning ◽  
Convolutional Neural Network ◽  
Test Data ◽  
Early Stage ◽  
Normal Subjects ◽  
Diagnostic Ability ◽  
Data Set ◽  
Sd Oct

Background/AimsTo evaluate, with spectral-domain optical coherence tomography (SD-OCT), the glaucoma-diagnostic ability of a deep-learning classifier.MethodsA total of 777 Cirrus high-definition SD-OCT image sets of the retinal nerve fibre layer (RNFL) and ganglion cell-inner plexiform layer (GCIPL) of 315 normal subjects, 219 patients with early-stage primary open-angle glaucoma (POAG) and 243 patients with moderate-to-severe-stage POAG were aggregated. The image sets were divided into a training data set (252 normal, 174 early POAG and 195 moderate-to-severe POAG) and a test data set (63 normal, 45 early POAG and 48 moderate-to-severe POAG). The visual geometry group (VGG16)-based dual-input convolutional neural network (DICNN) was adopted for the glaucoma diagnoses. Unlike other networks, the DICNN structure takes two images (both RNFL and GCIPL) as inputs. The glaucoma-diagnostic ability was computed according to both accuracy and area under the receiver operating characteristic curve (AUC).ResultsFor the test data set, DICNN could distinguish between patients with glaucoma and normal subjects accurately (accuracy=92.793%, AUC=0.957 (95% CI 0.943 to 0.966), sensitivity=0.896 (95% CI 0.896 to 0.917), specificity=0.952 (95% CI 0.921 to 0.952)). For distinguishing between patients with early-stage glaucoma and normal subjects, DICNN’s diagnostic ability (accuracy=85.185%, AUC=0.869 (95% CI 0.825 to 0.879), sensitivity=0.921 (95% CI 0.813 to 0.905), specificity=0.756 (95% CI 0.610 to 0.790)]) was higher than convolutional neural network algorithms that trained with RNFL or GCIPL separately.ConclusionThe deep-learning algorithm using SD-OCT can distinguish normal subjects not only from established patients with glaucoma but also from patients with early-stage glaucoma. The deep-learning model with DICNN, as trained by both RNFL and GCIPL thickness map data, showed a high diagnostic ability for discriminatingpatients with early-stage glaucoma from normal subjects.

OCTA characterisation of microvascular retinal alterations in patients with central serous chorioretinopathy

2020 ◽  
Vol 104 (10) ◽  
pp. 1453-1457 ◽  
Author(s):  
Marco Battista ◽  
Enrico Borrelli ◽  
Mariacristina Parravano ◽  
Francesco Gelormini ◽  
Massimiliano Tedeschi ◽  
...  
Keyword(s):  
Optical Coherence Tomography ◽  
Central Serous Chorioretinopathy ◽  
Optical Coherence Tomography Angiography ◽  
Spectral Domain ◽  
Foveal Avascular Zone ◽  
Optical Coherence ◽  
Vascular Changes ◽  
Chronic Central Serous Chorioretinopathy ◽  
Avascular Zone ◽  
Sd Oct

PurposeThis study aimed to describe the characteristics of microvascular retinal alterations in eyes with chronic central serous chorioretinopathy (CSC) employing optical coherence tomography angiography (OCTA) analysis.MethodsWe collected data from 472 eyes with chronic CSC from 336 patients who had OCTA obtained. Each OCTA image was graded by two readers to assess the presence of microvascular retinal alterations, including regions of vascular rarefaction/retinal hypoperfusion, enlargement of the foveal avascular zone (FAZ) and presence of telangiectasias or microaneurysms. Volume spectral domain optical coherence tomography (SD-OCT) scans were obtained through the macula and the OCT was correlated with the OCTA findings in eyes with retinal vascular alterations.ResultsOCTA displayed microvascular retinal alterations in 18 out of 474 eyes (3.6%) from 14 patients (13 male and 1 female; mean±SD age was 54.7±11.1 years). One eye displayed the presence of retinal telangiectasias, while 17 out of 18 eyes were graded as having areas of retinal vascular rarefactions, and 3 out of 17 eyes were also characterised by an enlargement of the FAZ. The parafoveal region was the location most involved by retinal vascular changes (66,7%), followed by foveal (22,2%) and perifoveal (11.1%) regions, respectively.ConclusionAlthough CSC is known to represent a choroidal disorder, retinal vascular alterations may be present in these eyes and OCTA may represent a useful tool to identify and describe them.

Sign in / Sign up

Export Citation Format

Share Document