scholarly journals Estimating Rates of Progression and Predicting Future Visual Fields in Glaucoma Using a Deep Variational Autoencoder

2019 ◽  
Vol 9 (1) ◽  
Author(s):  
Samuel I. Berchuck ◽  
Sayan Mukherjee ◽  
Felipe A. Medeiros
Keyword(s):  
Visual Field ◽  
Learning Algorithm ◽  
Visual Fields ◽  
Visual Field Loss ◽  
Absolute Error ◽  
Rate Of Change ◽  
Mean Deviation ◽  
Automated Perimetry ◽  
Standard Automated Perimetry ◽  
Deep Learning Algorithm

AbstractIn this manuscript we develop a deep learning algorithm to improve estimation of rates of progression and prediction of future patterns of visual field loss in glaucoma. A generalized variational auto-encoder (VAE) was trained to learn a low-dimensional representation of standard automated perimetry (SAP) visual fields using 29,161 fields from 3,832 patients. The VAE was trained on a 90% sample of the data, with randomization at the patient level. Using the remaining 10%, rates of progression and predictions were generated, with comparisons to SAP mean deviation (MD) rates and point-wise (PW) regression predictions, respectively. The longitudinal rate of change through the VAE latent space (e.g., with eight dimensions) detected a significantly higher proportion of progression than MD at two (25% vs. 9%) and four (35% vs 15%) years from baseline. Early on, VAE improved prediction over PW, with significantly smaller mean absolute error in predicting the 4th, 6th and 8th visits from the first three (e.g., visit eight: VAE8: 5.14 dB vs. PW: 8.07 dB; P < 0.001). A deep VAE can be used for assessing both rates and trajectories of progression in glaucoma, with the additional benefit of being a generative technique capable of predicting future patterns of visual field damage.

scholarly journals Estimating Rates of Progression and Predicting Future Visual Fields in Glaucoma Using a Deep Variational Autoencoder

2019 ◽  
Author(s):  
Samuel I. Berchuck ◽  
Sayan Mukherjee ◽  
Felipe A. Medeiros
Keyword(s):  
Visual Field ◽  
Learning Algorithm ◽  
Visual Fields ◽  
Visual Field Loss ◽  
Ordinary Least Squares ◽  
Rate Of Change ◽  
Feature Representation ◽  
Mean Deviation ◽  
Automated Perimetry ◽  
Standard Automated Perimetry

ABSTRACTPurposeTo develop a novel deep learning algorithm to improve estimation of rates of progression and prediction of future patterns of visual field loss in glaucoma.DesignProspective observational cohort.MethodsA variational auto-encoder (VAE) was trained to learn a low-dimensional feature representation of standard automated perimetry (SAP) visual fields using 29,161 fields from 3,832 patients. The VAE was trained on a 90% sample of the data, with randomization at the patient level. Using the remaining 10%, rates of progression and predictions were generated, with comparisons to SAP mean deviation (MD) rates and point-wise (PW) regression predictions, respectively. From the VAE, rates were calculated using the average of slopes across latent features from ordinary least squares (OLS) regression and trajectories of the features were used to generate predictions.ResultsThe longitudinal rate of change through the VAE latent space (e.g., with eight dimensions) detected a significantly higher proportion of progression than MD at two (19% vs. 6%) and four (40% vs 14%) years from baseline. Early on, VAE improved prediction over PW, with significantly smaller mean absolute error in predicting the 4th, 6th and 8th visits from the first three (e.g., visit eight: VAE8: 4.06 dB vs. PW: 6.06 dB; P<0.001).ConclusionA deep VAE can be used for assessing both rates and trajectories of progression in glaucoma, with the additional benefit of being a generative technique capable of predicting future patterns of visual field damage in the disease.

scholarly journals Estimating visual field loss from monoscopic optic disc photography using deep learning model

2020 ◽  
Vol 10 (1) ◽  
Author(s):  
Jinho Lee ◽  
Yong Woo Kim ◽  
Ahnul Ha ◽  
Young Kook Kim ◽  
Ki Ho Park ◽  
...  
Keyword(s):  
Deep Learning ◽  
Visual Field ◽  
Optic Disc ◽  
Correlation Coefficient ◽  
Characteristic Curve ◽  
Visual Field Loss ◽  
Absolute Error ◽  
Mean Deviation ◽  
Automated Perimetry ◽  
Standard Automated Perimetry

AbstractVisual field assessment is recognized as the important criterion of glaucomatous damage judgement; however, it can show large test–retest variability. We developed a deep learning (DL) algorithm that quantitatively predicts mean deviation (MD) of standard automated perimetry (SAP) from monoscopic optic disc photographs (ODPs). A total of 1200 image pairs (ODPs and SAP results) for 563 eyes of 327 participants were enrolled. A DL model was built by combining a pre-trained DL network and subsequently trained fully connected layers. The correlation coefficient and mean absolute error (MAE) between the predicted and measured MDs were calculated. The area under the receiver operating characteristic curve (AUC) was calculated to evaluate the detection ability for glaucomatous visual field (VF) loss. The data were split into training/validation (1000 images) and testing (200 images) sets to evaluate the performance of the algorithm. The predicted MD showed a strong correlation and good agreement with the actual MD (correlation coefficient = 0.755; R2 = 57.0%; MAE = 1.94 dB). The model also accurately predicted the presence of glaucomatous VF loss (AUC 0.953). The DL algorithm showed great feasibility for prediction of MD and detection of glaucomatous functional loss from ODPs.

scholarly journals Glaucoma home-monitoring using a tablet-based visual field test (Eyecatcher): An assessment of accuracy and adherence over six months

2020 ◽  
Author(s):  
Pete R. Jones ◽  
Peter Campbell ◽  
Tamsin Callaghan ◽  
Lee Jones ◽  
Daniel S. Asfaw ◽  
...  
Keyword(s):  
Visual Field ◽  
Home Monitoring ◽  
Absolute Error ◽  
Machine Learning Techniques ◽  
Mean Deviation ◽  
Test Duration ◽  
Automated Perimetry ◽  
Standard Automated Perimetry ◽  
Measurement Variability ◽  
Prospective Longitudinal

AbstractPurposeTo assess accuracy and adherence of visual field (VF) home-monitoring in a pilot sample of glaucoma patients.DesignProspective longitudinal observation.MethodsTwenty adults (median 71 years) with an established diagnosis of glaucoma were issued a tablet-perimeter (Eyecatcher), and were asked to perform one VF home-assessment per eye, per month, for 6 months (12 tests total). Before and after home-monitoring, two VF assessments were performed in-clinic using Standard Automated Perimetry (SAP; 4 tests total, per eye).ResultsAll 20 participants could perform monthly home-monitoring, though one participant stopped after 4 months (Adherence: 98%). There was good concordance between VFs measured at home and in the clinic (r = 0.94, P < 0.001). In 21 of 236 tests (9%) Mean Deviation deviated by more than ±3dB from the median. Many of these anomalous tests could be identified by applying machine learning techniques to recordings from the tablets’ front-facing camera (Area Under the ROC Curve = 0.78). Adding home-monitoring data to 2 SAP tests made 6 months apart reduced measurement error (between-test measurement variability) in 97% of eyes, with mean absolute error more than halving in 90% of eyes. Median test duration was 4.5mins (Quartiles: 3.9−5.2mins). Substantial variations in ambient illumination had no observable effect on VF measurements (r = 0.07, P = 0.320).ConclusionsHome-monitoring of VFs is viable for some patients, and may provide clinically useful data.

scholarly journals The effect of intraocular pressure on visual field after trabeculectomy in patients with primary open angle glaucoma

2012 ◽  
Vol 59 (1) ◽  
pp. 61-66
Author(s):  
Paraskeva Hentova-Sencanic ◽  
Marija Bozic ◽  
Ivan Sencanic ◽  
Milan Stojcic ◽  
Branislav Stankovic ◽  
...  
Keyword(s):  
Visual Field ◽  
Linear Regression Analysis ◽  
Regression Line ◽  
Visual Fields ◽  
Visual Field Loss ◽  
Rate Of Change ◽  
Negative Slope ◽  
Automated Perimetry ◽  
Percentage Reduction ◽  
The Mean

Purpose: To compare the mean intraocular pres-sure (IOP), peak IOP and percentage reduction in IOP in the first live years following trabeculectomy between the patients with progressed visual field loss and the patients with stable visual fields. Material and methods: Thirty-six eyes of 36 patients were followed for five years after their first trabeculectomy with tonometry and automated perimetry (Octopus 500EZ, program Gl). The rate of change of the visual field was measured by linear regression analysis of the mean sensitivity value (dB9) of each field test versus time (month). Based on the statistical significance of the slope of the regression line (Spear man p value of the correlation coefficient less than 0.05), patients were divided into two groups: with significant negative slope of the regression line (group with progressed visual field loss) and with non-significant slope of the regression line (group with stable visual field). The mean IOP values and percentage of IOP reduction at the end of each of the first live years after surgery were compared between the group with progressed field loss and group with stable fields by using Mann-Whitney U test. Results: Patients with progressed visual field loss had higher mean IOP, higher peak IOP and less reduction in pressure after the operation than patients with stable visual field. The mean IOP at end of the two year postoperative period was significantly higher in patients with progressed visual field loss (21.98?3.38mmHg) than in those with stable fields (17.48?4,80mmHg). The mean percentage reduction in IOP at the end of two year postoperative peri?od was significantly less in patients that showed progression of field loss (21.84%) than in those with stable fields (41.0%). Conclusion: Prognosis for further field loss seems to be better if postoperative pressure is at lower levels and greater percent reduction of IOP is obtained after surgery. The data that predict better prognosis is the mean postoperative IOP value of approximately 18mmHg or less resulting from at least 35% of IOP reduction.

scholarly journals Association between visual field damage and corneal structural parameters

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Alexandru Lavric ◽  
Valentin Popa ◽  
Hidenori Takahashi ◽  
Rossen M. Hazarbassanov ◽  
Siamak Yousefi
Keyword(s):  
Machine Learning ◽  
Visual Field ◽  
Structural Parameters ◽  
Corneal Thickness ◽  
Anterior Segment ◽  
Mean Deviation ◽  
Automated Perimetry ◽  
Standard Automated Perimetry ◽  
Visual Field Damage ◽  
Corneal Shape

AbstractThe main goal of this study is to identify the association between corneal shape, elevation, and thickness parameters and visual field damage using machine learning. A total of 676 eyes from 568 patients from the Jichi Medical University in Japan were included in this study. Corneal topography, pachymetry, and elevation images were obtained using anterior segment optical coherence tomography (OCT) and visual field tests were collected using standard automated perimetry with 24-2 Swedish Interactive Threshold Algorithm. The association between corneal structural parameters and visual field damage was investigated using machine learning and evaluated through tenfold cross-validation of the area under the receiver operating characteristic curves (AUC). The average mean deviation was − 8.0 dB and the average central corneal thickness (CCT) was 513.1 µm. Using ensemble machine learning bagged trees classifiers, we detected visual field abnormality from corneal parameters with an AUC of 0.83. Using a tree-based machine learning classifier, we detected four visual field severity levels from corneal parameters with an AUC of 0.74. Although CCT and corneal hysteresis have long been accepted as predictors of glaucoma development and future visual field loss, corneal shape and elevation parameters may also predict glaucoma-induced visual functional loss.

Compass fundus automated perimetry

2018 ◽  
Vol 28 (5) ◽  
pp. 481-490 ◽  
Author(s):  
Paolo Fogagnolo ◽  
Maurizio Digiuni ◽  
Giovanni Montesano ◽  
Chiara Rui ◽  
Marco Morales ◽  
...  
Keyword(s):  
Visual Field ◽  
Visual Fields ◽  
Field Testing ◽  
Automated Perimetry ◽  
Standard Automated Perimetry ◽  
Gaze Tracking ◽  
Intrinsic Factors ◽  
Global Indices ◽  
Humphrey Field Analyzer ◽  
The Impact

Background: Compass (CenterVue, Padova, Italy) is a fundus automated perimeter which has been introduced in the clinical practice for glaucoma management in 2014. The aim of the article is to review Compass literature, comparing its performances against Humphrey Field Analyzer (Zeiss Humphrey Systems, Dublin, CA, USA). Results: Analyses on both normal and glaucoma subjects agree on the fact that Humphrey Field Analyzer and Compass are interchangeable, as the difference of their global indices is largely inferior than test -retest variability for Humphrey Field Analyzer. Compass also enables interesting opportunities for the assessment of morphology, and the integration between morphology and function on the same device. Conclusion: Visual field testing by standard automated perimetry is limited by a series of intrinsic factors related to the psychophysical nature of the examination; recent papers suggest that gaze tracking is closely related to visual field reliability. Compass, thanks to a retinal tracker and to the active dislocation of stimuli to compensate for eye movements, is able to provide visual fields unaffected by fixation instability. Also, the instrument is a true colour, confocal retinoscope and obtains high-quality 60° × 60° photos of the central retina and stereo-photos details of the optic nerve. Overlapping the image of the retina to field sensitivity may be useful in ascertaining the impact of comorbidities. In addition, the recent introduction of stereoscopic photography may be very useful for better clinical examination.

Detecting Visual Field Progression in Glaucoma – Using the Right Tools for the Job

2013 ◽  
Vol 07 (01) ◽  
pp. 20 ◽  
Author(s):  
Luke J Saunders ◽  
Richard A Russell ◽  
David P Crabb ◽  
◽  
◽  
...  
Keyword(s):  
Visual Field ◽  
Visual Fields ◽  
Field Measurements ◽  
Automated Perimetry ◽  
Standard Automated Perimetry ◽  
Visual Disability ◽  
New Research ◽  
The Right ◽  
Over Time

Monitoring disease progression is at the centre of managing a patient with glaucoma. This article focuses specifically on how visual field measurements from standard automated perimetry (SAP) can be monitored over time. Various options for analysis on the Humphrey and Octopus perimeters are discussed, from summary indices to event and trend-based analyses; their respective merits and flaws evaluated. It is strongly recommended that quantitative analysis methods and software are used in assessing progression, as variability in threshold measurements makes detecting true deterioration non-trivial. Recommendations on the frequency of visual fields that should be taken per year are also discussed. The article concludes by putting the spotlight on new research being undertaken to improve the methods of measuring and predicting progression, as well as relating visual fields to patient visual disability and quality of life.

scholarly journals Predicting the central 10 degrees visual field in glaucoma by applying a deep learning algorithm to optical coherence tomography images

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Shotaro Asano ◽  
Ryo Asaoka ◽  
Hiroshi Murata ◽  
Yohei Hashimoto ◽  
Atsuya Miki ◽  
...  
Keyword(s):  
Optical Coherence Tomography ◽  
Visual Field ◽  
Learning Algorithm ◽  
Normal Subjects ◽  
Absolute Error ◽  
Test Point ◽  
Training Dataset ◽  
Optical Coherence ◽  
Deep Learning Algorithm ◽  
Testing Dataset

AbstractWe aimed to develop a model to predict visual field (VF) in the central 10 degrees in patients with glaucoma, by training a convolutional neural network (CNN) with optical coherence tomography (OCT) images and adjusting the values with Humphrey Field Analyzer (HFA) 24–2 test. The training dataset included 558 eyes from 312 glaucoma patients and 90 eyes from 46 normal subjects. The testing dataset included 105 eyes from 72 glaucoma patients. All eyes were analyzed by the HFA 10-2 test and OCT; eyes in the testing dataset were additionally analyzed by the HFA 24-2 test. During CNN model training, the total deviation (TD) values of the HFA 10-2 test point were predicted from the combined OCT-measured macular retinal layers’ thicknesses. Then, the predicted TD values were corrected using the TD values of the innermost four points from the HFA 24-2 test. Mean absolute error derived from the CNN models ranged between 9.4 and 9.5 B. These values reduced to 5.5 dB on average, when the data were corrected using the HFA 24-2 test. In conclusion, HFA 10-2 test results can be predicted with a OCT images using a trained CNN model with adjustment using HFA 24-2 test.

Sign in / Sign up

Export Citation Format

Share Document