Automated detection of diabetic retinopathy using custom convolutional neural network

2022 ◽  
pp. 1-17
Author(s):  
Saleh Albahli ◽  
Ghulam Nabi Ahmad Hassan Yar
Keyword(s):  
Diabetic Retinopathy ◽  
Deep Learning ◽  
Macular Edema ◽  
Training Data ◽  
Validation Dataset ◽  
Learning Models ◽  
Validation Data ◽  
Study Results ◽  
Image Dataset ◽  
Different Types

Diabetic retinopathy is an eye deficiency that affects retina as a result of the patient having diabetes mellitus caused by high sugar levels, which may eventually lead to macular edema. The objective of this study is to design and compare several deep learning models that detect severity of diabetic retinopathy, determine risk of leading to macular edema, and segment different types of disease patterns using retina images. Indian Diabetic Retinopathy Image Dataset (IDRiD) dataset was used for disease grading and segmentation. Since images of the dataset have different brightness and contrast, we employed three techniques for generating processed images from the original images, which include brightness, color and, contrast (BCC) enhancing, color jitters (CJ), and contrast limited adaptive histogram equalization (CLAHE). After image preporcessing, we used pre-trained ResNet50, VGG16, and VGG19 models on these different preprocessed images both for determining the severity of the retinopathy and also the chances of macular edema. UNet was also applied to segment different types of diseases. To train and test these models, image dataset was divided into training, testing, and validation data at 70%, 20%, and 10% ratios, respectively. During model training, data augmentation method was also applied to increase the number of training images. Study results show that for detecting the severity of retinopathy and macular edema, ResNet50 showed the best accuracy using BCC and original images with an accuracy of 60.2% and 82.5%, respectively, on validation dataset. In segmenting different types of diseases, UNet yielded the highest testing accuracy of 65.22% and 91.09% for microaneurysms and hard exudates using BCC images, 84.83% for optic disc using CJ images, 59.35% and 89.69% for hemorrhages and soft exudates using CLAHE images, respectively. Thus, image preprocessing can play an important role to improve efficacy and performance of deep learning models.

scholarly journals Domain randomization-enhanced deep learning models for bird detection

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Xin Mao ◽  
Jun Kang Chow ◽  
Pin Siang Tan ◽  
Kuan-fu Liu ◽  
Jimmy Wu ◽  
...  
Keyword(s):  
Deep Learning ◽  
Continuous Monitoring ◽  
Bird Species ◽  
Training Data ◽  
Learning Models ◽  
Fine Grained ◽  
Bird Detection ◽  
Relationship Of ◽  
The Relationship

AbstractAutomatic bird detection in ornithological analyses is limited by the accuracy of existing models, due to the lack of training data and the difficulties in extracting the fine-grained features required to distinguish bird species. Here we apply the domain randomization strategy to enhance the accuracy of the deep learning models in bird detection. Trained with virtual birds of sufficient variations in different environments, the model tends to focus on the fine-grained features of birds and achieves higher accuracies. Based on the 100 terabytes of 2-month continuous monitoring data of egrets, our results cover the findings using conventional manual observations, e.g., vertical stratification of egrets according to body size, and also open up opportunities of long-term bird surveys requiring intensive monitoring that is impractical using conventional methods, e.g., the weather influences on egrets, and the relationship of the migration schedules between the great egrets and little egrets.

scholarly journals U-Infuse: Democratization of Customizable Deep Learning for Object Detection

Sensors ◽  
2021 ◽  
Vol 21 (8) ◽  
pp. 2611
Author(s):  
Andrew Shepley ◽  
Greg Falzon ◽  
Christopher Lawson ◽  
Paul Meek ◽  
Paul Kwan
Keyword(s):  
Deep Learning ◽  
Intellectual Property ◽  
Object Detection ◽  
Image Data ◽  
Learning Technologies ◽  
Training Data ◽  
Learning Models ◽  
Ecological Data ◽  
Single Class ◽  
Large Numbers

Image data is one of the primary sources of ecological data used in biodiversity conservation and management worldwide. However, classifying and interpreting large numbers of images is time and resource expensive, particularly in the context of camera trapping. Deep learning models have been used to achieve this task but are often not suited to specific applications due to their inability to generalise to new environments and inconsistent performance. Models need to be developed for specific species cohorts and environments, but the technical skills required to achieve this are a key barrier to the accessibility of this technology to ecologists. Thus, there is a strong need to democratize access to deep learning technologies by providing an easy-to-use software application allowing non-technical users to train custom object detectors. U-Infuse addresses this issue by providing ecologists with the ability to train customised models using publicly available images and/or their own images without specific technical expertise. Auto-annotation and annotation editing functionalities minimize the constraints of manually annotating and pre-processing large numbers of images. U-Infuse is a free and open-source software solution that supports both multiclass and single class training and object detection, allowing ecologists to access deep learning technologies usually only available to computer scientists, on their own device, customised for their application, without sharing intellectual property or sensitive data. It provides ecological practitioners with the ability to (i) easily achieve object detection within a user-friendly GUI, generating a species distribution report, and other useful statistics, (ii) custom train deep learning models using publicly available and custom training data, (iii) achieve supervised auto-annotation of images for further training, with the benefit of editing annotations to ensure quality datasets. Broad adoption of U-Infuse by ecological practitioners will improve ecological image analysis and processing by allowing significantly more image data to be processed with minimal expenditure of time and resources, particularly for camera trap images. Ease of training and use of transfer learning means domain-specific models can be trained rapidly, and frequently updated without the need for computer science expertise, or data sharing, protecting intellectual property and privacy.

scholarly journals Foveal avascular zone segmentation in optical coherence tomography angiography images using a deep learning approach

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Reza Mirshahi ◽  
Pasha Anvari ◽  
Hamid Riazi-Esfahani ◽  
Mahsa Sardarinia ◽  
Masood Naseripour ◽  
...  
Keyword(s):  
Optical Coherence Tomography ◽  
Diabetic Retinopathy ◽  
Deep Learning ◽  
Healthy Subjects ◽  
Optical Coherence Tomography Angiography ◽  
Validation Dataset ◽  
Diabetic Patients ◽  
Dice Similarity Coefficient ◽  
Optical Coherence ◽  
Avascular Zone

AbstractThe purpose of this study was to introduce a new deep learning (DL) model for segmentation of the fovea avascular zone (FAZ) in en face optical coherence tomography angiography (OCTA) and compare the results with those of the device’s built-in software and manual measurements in healthy subjects and diabetic patients. In this retrospective study, FAZ borders were delineated in the inner retinal slab of 3 × 3 enface OCTA images of 131 eyes of 88 diabetic patients and 32 eyes of 18 healthy subjects. To train a deep convolutional neural network (CNN) model, 126 enface OCTA images (104 eyes with diabetic retinopathy and 22 normal eyes) were used as training/validation dataset. Then, the accuracy of the model was evaluated using a dataset consisting of OCTA images of 10 normal eyes and 27 eyes with diabetic retinopathy. The CNN model was based on Detectron2, an open-source modular object detection library. In addition, automated FAZ measurements were conducted using the device’s built-in commercial software, and manual FAZ delineation was performed using ImageJ software. Bland–Altman analysis was used to show 95% limit of agreement (95% LoA) between different methods. The mean dice similarity coefficient of the DL model was 0.94 ± 0.04 in the testing dataset. There was excellent agreement between automated, DL model and manual measurements of FAZ in healthy subjects (95% LoA of − 0.005 to 0.026 mm2 between automated and manual measurement and 0.000 to 0.009 mm2 between DL and manual FAZ area). In diabetic eyes, the agreement between DL and manual measurements was excellent (95% LoA of − 0.063 to 0.095), however, there was a poor agreement between the automated and manual method (95% LoA of − 0.186 to 0.331). The presence of diabetic macular edema and intraretinal cysts at the fovea were associated with erroneous FAZ measurements by the device’s built-in software. In conclusion, the DL model showed an excellent accuracy in detection of FAZ border in enfaces OCTA images of both diabetic patients and healthy subjects. The DL and manual measurements outperformed the automated measurements of the built-in software.

Fully automated contrast and non-contrast cardiac view detection in echocardiography a multi-centre, multi-vendor study

2020 ◽  
Vol 41 (Supplement_2) ◽  
Author(s):  
S Gao ◽  
D Stojanovski ◽  
A Parker ◽  
P Marques ◽  
S Heitner ◽  
...  
Keyword(s):  
Neural Network ◽  
Training Data ◽  
Classification Model ◽  
Validation Dataset ◽  
Funding Source ◽  
Private Company ◽  
Validation Data ◽  
Independent Test ◽  
Model Training ◽  
Confusion Matrices

Abstract Background Correctly identifying views acquired in a 2D echocardiographic examination is paramount to post-processing and quantification steps often performed as part of most clinical workflows. In many exams, particularly in stress echocardiography, microbubble contrast is used which greatly affects the appearance of the cardiac views. Here we present a bespoke, fully automated convolutional neural network (CNN) which identifies apical 2, 3, and 4 chamber, and short axis (SAX) views acquired with and without contrast. The CNN was tested in a completely independent, external dataset with the data acquired in a different country than that used to train the neural network. Methods Training data comprised of 2D echocardiograms was taken from 1014 subjects from a prospective multisite, multi-vendor, UK trial with the number of frames in each view greater than 17,500. Prior to view classification model training, images were processed using standard techniques to ensure homogenous and normalised image inputs to the training pipeline. A bespoke CNN was built using the minimum number of convolutional layers required with batch normalisation, and including dropout for reducing overfitting. Before processing, the data was split into 90% for model training (211,958 frames), and 10% used as a validation dataset (23,946 frames). Image frames from different subjects were separated out entirely amongst the training and validation datasets. Further, a separate trial dataset of 240 studies acquired in the USA was used as an independent test dataset (39,401 frames). Results Figure 1 shows the confusion matrices for both validation data (left) and independent test data (right), with an overall accuracy of 96% and 95% for the validation and test datasets respectively. The accuracy for the non-contrast cardiac views of >99% exceeds that seen in other works. The combined datasets included images acquired across ultrasound manufacturers and models from 12 clinical sites. Conclusion We have developed a CNN capable of automatically accurately identifying all relevant cardiac views used in “real world” echo exams, including views acquired with contrast. Use of the CNN in a routine clinical workflow could improve efficiency of quantification steps performed after image acquisition. This was tested on an independent dataset acquired in a different country to that used to train the model and was found to perform similarly thus indicating the generalisability of the model. Figure 1. Confusion matrices Funding Acknowledgement Type of funding source: Private company. Main funding source(s): Ultromics Ltd.

scholarly journals A Physics-Infused Deep Learning Model for the Prediction of Refractive Indices and Its Use for the Large-Scale Screening of Organic Compound Space

2019 ◽  
Author(s):  
Mojtaba Haghighatlari ◽  
Gaurav Vishwakarma ◽  
Mohammad Atif Faiz Afzal ◽  
Johannes Hachmann
Keyword(s):  
Machine Learning ◽  
Deep Learning ◽  
Large Scale ◽  
Organic Molecules ◽  
Learning Model ◽  
Training Data ◽  
Refractive Indices ◽  
Learning Models ◽  
Deep Learning Model ◽  
Machine Learning Models

<div><div><div><p>We present a multitask, physics-infused deep learning model to accurately and efficiently predict refractive indices (RIs) of organic molecules, and we apply it to a library of 1.5 million compounds. We show that it outperforms earlier machine learning models by a significant margin, and that incorporating known physics into data-derived models provides valuable guardrails. Using a transfer learning approach, we augment the model to reproduce results consistent with higher-level computational chemistry training data, but with a considerably reduced number of corresponding calculations. Prediction errors of machine learning models are typically smallest for commonly observed target property values, consistent with the distribution of the training data. However, since our goal is to identify candidates with unusually large RI values, we propose a strategy to boost the performance of our model in the remoter areas of the RI distribution: We bias the model with respect to the under-represented classes of molecules that have values in the high-RI regime. By adopting a metric popular in web search engines, we evaluate our effectiveness in ranking top candidates. We confirm that the models developed in this study can reliably predict the RIs of the top 1,000 compounds, and are thus able to capture their ranking. We believe that this is the first study to develop a data-derived model that ensures the reliability of RI predictions by model augmentation in the extrapolation region on such a large scale. These results underscore the tremendous potential of machine learning in facilitating molecular (hyper)screening approaches on a massive scale and in accelerating the discovery of new compounds and materials, such as organic molecules with high-RI for applications in opto-electronics.</p></div></div></div>

RECONSTRUCTED TEETH IMAGE FROM BRACES WITH GAN

2021 ◽  
pp. 2150043
Author(s):  
Vu Tuan Hai ◽  
Dang Thanh Vu ◽  
Huynh Ho Thi Mong Trinh ◽  
Pham The Bao
Keyword(s):  
Health Care ◽  
Deep Learning ◽  
Training Data ◽  
Learning Models ◽  
Paired Data ◽  
Object Removal ◽  
Interactive Tool ◽  
Learning Techniques ◽  
Large Context ◽  
Image Translation

Recent advances in deep learning models have shown promising potential in object removal, which is the task of replacing undesired objects with appropriate pixel values using known context. Object removal-based deep learning can commonly be solved by modeling it as the Img2Img (image to image) translation or Inpainting. Instead of dealing with a large context, this paper aims at a specific application of object removal, that is, erasing braces trace out of an image having teeth with braces (called braces2teeth problem). We solved the problem by three methods corresponding to different datasets. Firstly, we use the CycleGAN model to deal with the problem that paired training data is not available. In the second case, we try to create pseudo-paired data to train the Pix2Pix model. In the last case, we utilize GraphCut combining generative inpainting model to build a user-interactive tool that can improve the result in case the user is not satisfied with previous results. To our best knowledge, this study is one of the first attempts to take the braces2teeth problem into account by using deep learning techniques and it can be applied in various fields, from health care to entertainment.

Retinal blood vessels detection for diabetic retinopathy with Ridgelet transform and convolution neural network

2020 ◽  
Vol 18 (06) ◽  
pp. 2050048
Author(s):  
Nirmal Yadav
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Diabetic Retinopathy ◽  
Deep Learning ◽  
Blood Vessels ◽  
Machine Learning Algorithms ◽  
Convolution Neural Network ◽  
Transform Method ◽  
Ridgelet Transform ◽  
Study Results

Applying machine learning in life sciences, especially diagnostics, has become a key area of focus for researchers. Combining machine learning with traditional algorithms provides a unique opportunity of providing better solutions for the patients. In this paper, we present study results of applying the Ridgelet Transform method on retina images to enhance the blood vessels, then using machine learning algorithms to identify cases of Diabetic Retinopathy (DR). The Ridgelet transform provides better results for line singularity of image function and, thus, helps to reduce artefacts along the edges of the image. The Ridgelet Transform method, when compared with earlier known methods of image enhancement, such as Wavelet Transform and Contourlet Transform, provided satisfactory results. The transformed image using the Ridgelet Transform method with pre-processing quantifies the amount of information in the dataset. It efficiently enhances the generation of features vectors in the convolution neural network (CNN). In this study, a sample of fundus photographs was processed, which was obtained from a publicly available dataset. In pre-processing, first, CLAHE was applied, followed by filtering and application of Ridgelet transform on the patches to improve the quality of the image. Then, this processed image was used for statistical feature detection and classified by deep learning method to detect DR images from the dataset. The successful classification ratio was 98.61%. This result concludes that the transformed image of fundus using the Ridgelet Transform enables better detection by leveraging a transform-based algorithm and the deep learning.

Diagnostic assessment of a deep learning system for detecting atrial fibrillation in pulse waveforms

Heart ◽  
2018 ◽  
Vol 104 (23) ◽  
pp. 1921-1928 ◽  
Author(s):  
Ming-Zher Poh ◽  
Yukkee Cheung Poh ◽  
Pak-Hei Chan ◽  
Chun-Ka Wong ◽  
Louise Pun ◽  
...  
Keyword(s):  
Atrial Fibrillation ◽  
Deep Learning ◽  
Test Data ◽  
Predictive Value ◽  
Characteristic Curve ◽  
Performance Comparison ◽  
Learning System ◽  
Training Data ◽  
Validation Data ◽  
Data Set

ObjectiveTo evaluate the diagnostic performance of a deep learning system for automated detection of atrial fibrillation (AF) in photoplethysmographic (PPG) pulse waveforms.MethodsWe trained a deep convolutional neural network (DCNN) to detect AF in 17 s PPG waveforms using a training data set of 149 048 PPG waveforms constructed from several publicly available PPG databases. The DCNN was validated using an independent test data set of 3039 smartphone-acquired PPG waveforms from adults at high risk of AF at a general outpatient clinic against ECG tracings reviewed by two cardiologists. Six established AF detectors based on handcrafted features were evaluated on the same test data set for performance comparison.ResultsIn the validation data set (3039 PPG waveforms) consisting of three sequential PPG waveforms from 1013 participants (mean (SD) age, 68.4 (12.2) years; 46.8% men), the prevalence of AF was 2.8%. The area under the receiver operating characteristic curve (AUC) of the DCNN for AF detection was 0.997 (95% CI 0.996 to 0.999) and was significantly higher than all the other AF detectors (AUC range: 0.924–0.985). The sensitivity of the DCNN was 95.2% (95% CI 88.3% to 98.7%), specificity was 99.0% (95% CI 98.6% to 99.3%), positive predictive value (PPV) was 72.7% (95% CI 65.1% to 79.3%) and negative predictive value (NPV) was 99.9% (95% CI 99.7% to 100%) using a single 17 s PPG waveform. Using the three sequential PPG waveforms in combination (<1 min in total), the sensitivity was 100.0% (95% CI 87.7% to 100%), specificity was 99.6% (95% CI 99.0% to 99.9%), PPV was 87.5% (95% CI 72.5% to 94.9%) and NPV was 100% (95% CI 99.4% to 100%).ConclusionsIn this evaluation of PPG waveforms from adults screened for AF in a real-world primary care setting, the DCNN had high sensitivity, specificity, PPV and NPV for detecting AF, outperforming other state-of-the-art methods based on handcrafted features.

scholarly journals Essentials of a Robust Deep Learning System for Diabetic Retinopathy Screening: A Systematic Literature Review

2020 ◽  
Vol 2020 ◽  
pp. 1-11
Author(s):  
Aan Chu ◽  
David Squirrell ◽  
Andelka M. Phillips ◽  
Ehsan Vaghefi
Keyword(s):  
Diabetic Retinopathy ◽  
Deep Learning ◽  
Learning Algorithm ◽  
Learning Algorithms ◽  
Validation Dataset ◽  
Cochrane Library ◽  
Clinical Implementation ◽  
Diabetic Retinopathy Screening ◽  
Deep Learning Algorithm ◽  
Retinopathy Screening

This systematic review was performed to identify the specifics of an optimal diabetic retinopathy deep learning algorithm, by identifying the best exemplar research studies of the field, whilst highlighting potential barriers to clinical implementation of such an algorithm. Searching five electronic databases (Embase, MEDLINE, Scopus, PubMed, and the Cochrane Library) returned 747 unique records on 20 December 2019. Predetermined inclusion and exclusion criteria were applied to the search results, resulting in 15 highest-quality publications. A manual search through the reference lists of relevant review articles found from the database search was conducted, yielding no additional records. A validation dataset of the trained deep learning algorithms was used for creating a set of optimal properties for an ideal diabetic retinopathy classification algorithm. Potential limitations to the clinical implementation of such systems were identified as lack of generalizability, limited screening scope, and data sovereignty issues. It is concluded that deep learning algorithms in the context of diabetic retinopathy screening have reported impressive results. Despite this, the potential sources of limitations in such systems must be evaluated carefully. An ideal deep learning algorithm should be clinic-, clinician-, and camera-agnostic; complying with the local regulation for data sovereignty, storage, privacy, and reporting; whilst requiring minimum human input.

Sign in / Sign up

Export Citation Format

Share Document