Automated detection of diabetic retinopathy using convolutional neural networks on a small dataset

doi:10.1016/j.patrec.2020.04.026

Pattern Recognition Letters

Volume 135, July 2020, Pages 293-298

https://doi.org/10.1016/j.patrec.2020.04.026 Get rights and content

Highlights

•
Early detection of Diabetes-1 and Diabetes-2 from the transfer learning-based CNN architecture on color fundus photography.
•
Overfitting issue in CNN is resolved using the transfer learning technique and hence small datasets can be used for analysis.
•
A contrast enhancement pre-processing technique is used to preserve the brightness for effective analysis.

Abstract

Diabetic Retinopathy is a complication based on patients suffering from type-1 or type-2 diabetes. Early detection is essential as complication can lead to vision problems such as retinal detachment, vitreous hemorrhage and glaucoma. The principal stages of diabetic retinopathy are non-Proliferative diabetic retinopathy and Proliferative diabetic retinopathy. In this paper, we propose a transfer learning based CNN architecture on colour fundus photography that performs relatively well on a much smaller dataset of skewed classes of 3050 training images and 419 validation images in recognizing classes of Diabetic Retinopathy from hard exudates, blood vessels and texture. This model is extremely robust and lightweight, garnering a potential to work considerably well in small real time applications with limited computing power to speed up the screening process. The dataset was trained on Google Colab. We trained our model on 4 classes - I)No DR ii)Mild DR iii)Moderate DR iv)Proliferative DR, and achieved a Cohens Kappa score of 0.8836 on the validation set along with 0.9809 on the training set.

Introduction

Deep learning has enhanced the purpose of computer vision in identifying and classifying images and are a key tool used to automate tasks in our daily lives. Convolutional networks have been consistently developed for object detection, classification, segmentation. The use of convolutional neural networks (CNNs) on medical images has helped the medical sector immensely due to it's ability to learn representations of data [1].

Diabetic Retinopathy turns out to be a major cause of blindness in the western world, and regular screening of the patients reduces the risk of blindness. There are a number of features pertaining to the recognition of retinopathy in fundus photography, and computer vision based trained classifiers work pretty well in classification. Promising work has been displayed in the detection of retinopathy using k-NN classifiers and vector machines. CNNs have also been used for the classification of Diabetic Retinopathy [13,15], given a big dataset and considerable computing power [18]. They have been instrumental in detecting the features such as haemorrhage and hard exudes that identify retinopathy [8]. Deep architectures of CNN have been instrumental in providing the finesse and high performance to trained models by learning patterns from raw images [2]. Due to the availability of annotated data and evolution of GPUs, CNNs have been increasingly applicable in a number of cases. However, in case of medical datasets, huge amounts of annotated data are not readily available yet as the ImageNet Large Scale Visual Recognition Challenge (ILSVRC). Thus, transfer learning has not been very useful for medical datasets as most networks have been trained well to recognize objects present in the ImageNet dataset as shown in Fig. Fig. 1 and Fig. 5, Fig. 6, Fig. 7, Fig. 8, Fig. 9, Fig. 10, Fig. 11, Fig. 12, Fig. 13.

A major problem faced in training the model on less data is underfitting. Moreover, the presence of skewed classes causes the model to overfit on the largest class, which is turn, decreases the the corresponding F1 scores and Cohen's Kappa. Large datasets can often be over-sampled on the lower class, but oversampling on a small dataset will not be of much help against overfitting.

In this paper, we propose a deep learning based CNN method to classify images from a small and skewed dataset of 3050 training images belonging to 4 classes and 419 validation images to achieve a considerably good result. The accuracy metric used by us is Cohen's Kappa.

Section snippets

Related work

There has been considerable work regarding binary classification of Diabetic Retinopathy. Gardner et al used neural networks on 200 images by splitting an image into patches to achieve a sensitivity of 88.4% and specificity of 83.5% for binary classification. His work had been aided by a clinician in classifying the patches prior to using SVM.

Dr.Nayak et al also used neural networks to classify Diabetic Retinopathy Based on 3 classes by recognizing blood vessels and hard exudes from 140 images

Method and structure

We decided upon our network after studying baseline literature [7,22] and testing the performance of other models [[3], [4], [5],9,23]. It was observed that deeper layers cause overfitting as our dataset was comparatively smaller. In our network, we used CNN-based transfer learning on the DenseNet model pre-trained on ImageNet.

Results

The model was validated on 419 fundus images. The validation process was fast. We obtained a validation accuracy of 84.10 %. Accuracy was not used by us as the final metric due to the skewness of data. We chose Cohen's Kappa as our final metric than F1 score as Kappa provides a relativistic accuracy with respect to the other nearby classes, thus imparting a sense a reliability and originality in identification in case of a medical diagnosis [19]. ${Cohen}^{'} s Kappa (κ) = \frac{p_{o} - p_{e}}{1 - p_{e}} = 1 - \frac{1 - p_{o}}{1 - p_{e}}$

A literature

Discussion and conclusion

Our model has approached the classification of 4-class problem in Diabetic Retinopathy on a small dataset using deep learning. Most earlier algorithms dedicated to the classification of DR fundus images on a small dataset evaded the use of deep learning. Our method has produced comparable results with previous literature given data and hardware constraints and the presence of skewed classes. Transfer learning and fine-tuning on the pre-trained DenseNet has proved to be extremely effective on

Declaration of Competing Interest

None.

References (23)

M.D. Abramoff et al.
Automated early detection `of diabetic retinopathy
Ophthalmology
(2010)
Yann Lecun et al.
Deep learning
Nature
(2015)
Christian Szegedy, Wei Liu, Yangqing Jia, Pierre Sermanet, Scott Reed, Dragomir Anguelov, Dumitru Erhan, Vincent...
J. Nayak et al.
Automated identification of diabetic retinopathy stages using digital fundus images
J. Med. Syst.
(2008)
R. Acharya et al.
Application of higher order spectra for the identification of diabetes retinopathy stages
J. Med. Syst.
(2008)
P. Adarsh et al.
Multiclass svm-based automated diagnosis of diabetic retinopathy
Harry Pratt, Frans Coenen, Deborah M. Broadbent, Simon P. Harding, Yalin Zheng, Convolutional neural networks for...
Xiaogang Li, Tiantian Pang, Biao Xiong, Weixiang Liu, Ping Liang, Tianfu Wang, Convolutional neural networks based...
O. Faust et al.
Algorithms for the automated detection of diabetic retinopathy using digital fundus images: a review
J. Med. Syst.
(2012)
Carson Lam, Darvin Yi, Margaret Guo, Tony Lindsay, Automated detection of diabetic retinopathy using deep learning,...

...

Cited by (86)

NIMEQ-SACNet: A novel self-attention precision medicine model for vision-threatening diabetic retinopathy using image data
2024, Computers in Biology and Medicine
In the realm of precision medicine, the potential of deep learning is progressively harnessed to facilitate intricate clinical decision-making, especially when navigating multifaceted datasets encompassing Omics, Clinical, image, device, social, and environmental dimensions. This study accentuates the criticality of image data, given its instrumental role in detecting and classifying vision-threatening diabetic retinopathy (VTDR) – a predominant global contributor to vision impairment. The timely identification of VTDR is a linchpin for efficacious interventions and the mitigation of vision loss. Addressing this, This study introduces “NIMEQ-SACNet,” a novel hybrid model by the prowess of the Enhanced Quantum-Inspired Binary Grey Wolf Optimizer (EQI-BGWO) with a self-attention capsule network. The proposed approach is characterized by two pivotal advancements: firstly, the augmentation of the Binary Grey Wolf Optimization through Quantum Computing methodologies, and secondly, the deployment of the enhanced EQI-BGWO to adeptly calibrate the SACNet's parameters, culminating in a notable uplift in VTDR classification accuracy. The proposed model's ability to handle binary, 5-stage, and 7-stage VTDR classifications adroitly is noteworthy. Rigorous assessments on the fundus image dataset, underscored by metrics such as Accuracy, Sensitivity, Specificity, Precision, F1-Score, and MCC, bear testament to NIMEQ-SACNet's pre-eminence over prevailing algorithms and classification frameworks.
A refined ResNet18 architecture with Swish activation function for Diabetic Retinopathy classification
2024, Biomedical Signal Processing and Control
Automatic detection of Diabetic Retinopathy (DR) is critically important, as it is the primary reason of irreversible loss of vision in the economically active populations in the developed countries. Early detection of the onset of Diabetic Retinopathy can greatly benefit clinical treatment; although several different feature extraction methods have been proposed, the task of retinal image classification remains tedious even for trained clinicians. This paper emphasizes on Diabetic Retinopathy detection as well as the analysis of the different stages of DR, performed on fundus images using Deep Learning algorithms. Fundus images of the patient were provided as input to the developed model evaluated using the real-time dataset of the hospital. The proposed ResNet-18 architecture with swish function has achieved an accuracy of 93.51%, sensitivity of 93.42%, precision of 93.77% and F1-score of 93.59%. The paper concludes with a comparative study of Simple CNN, VGGNet-16, MobileNet-V2 and ResNet architectures and other state-of-art approaches, which highlights ResNet-18 with Swish as the most effective deep learning classifier model for DR detection.
From data to insights
2024, A Biologist's Guide to Artificial Intelligence: Building the foundations of Artificial Intelligence and Machine Learning for Achieving Advancements in Life Sciences
Managing diabetes is a complex task that requires continuous monitoring, personalized treatment plans, and informed decision-making. In recent years, advancements in machine learning (ML) have shown great potential in improving diabetes management by analyzing vast amounts of patient data and extracting valuable insights. This chapter presents a comprehensive overview of the utilization of the ML techniques in diabetes management, focusing on the journey from data collection to generating actionable insights. This chapter discusses the various sources of data available for diabetes management, including electronic health records, continuous glucose monitoring systems, wearable devices, and patient-reported data. The chapter delves into the application of ML algorithms in diabetes management. It examines various ML techniques such as supervised learning, unsupervised learning, and deep learning and discusses their specific applications in predicting blood glucose levels and identifying patterns and trends. Furthermore, the chapter explores the interpretability and explainability of ML models in diabetes management. It emphasizes the importance of transparent models that can provide meaningful insights to healthcare professionals and patients, enabling them to understand the underlying factors contributing to diabetes progression and treatment outcomes.
Diabetic retinopathy grading review: Current techniques and future directions
2023, Image and Vision Computing
Diabetic retinopathy (DR) is widely recognized as a leading cause of blindness among individuals with diabetes worldwide. Therefore, early diagnosis of DR plays a crucial role in preserving patients' vision and halting the progression of the disease to advanced stages. However, manual diagnosis of DR in clinical practice is time-consuming and susceptible to human error, especially during the early stages when the lesions associated with DR are often minute and challenging to identify. Furthermore, with the projected surge in the number of diabetic patients and a concurrent shortage of ophthalmologists, there will be insufficient healthcare professionals available to examine all individuals at risk. The application of machine- and deep learning-based techniques has proven effective in diagnosing medical images, including those related to DR. In this review, we surveyed and analyzed 55 DR grading studies published between 2018 and 2022 extracted from four academic digital libraries: Scopus, Web of Science, Google Scholar, and Science Direct. The review provides a comprehensive discussion and analysis of these selected studies, considering various aspects such as benchmark DR datasets, classification tasks, preprocessing techniques, learning approaches, and performance evaluation measures. Within the literature on DR grading, supervised-based learning techniques have been found to be more prevalent than semi-supervised learning techniques. Furthermore, researchers predominantly addressed this problem as an image-level classification task, overlooking the distinctive characteristics of lesions within each grade. Numerous proposed techniques primarily concentrate on detecting the early stages of DR, while a limited number of studies address the disease's advanced stages. The primary findings of our analysis reveal a potential direction for future research that emphasizes data- and model-centric approaches.
Diabetic retinopathy identification using parallel convolutional neural network based feature extractor and ELM classifier
2023, Expert Systems with Applications
Diabetic retinopathy (DR) is an incurable retinal condition caused by excessive blood sugar that, if left untreated, can result in even blindness. A novel automated technique for DR detection has been proposed in this paper. To accentuate the lesions, the fundus images (FIs) were preprocessed using Contrast Limited Adaptive Histogram Equalization (CLAHE). A parallel convolutional neural network (PCNN) was employed for feature extraction and then the extreme learning machine (ELM) technique was utilized for the DR classification. In comparison to the similar CNN structure, the PCNN design uses fewer parameters and layers, which minimizes the time required to extract distinctive features. The effectiveness of the technique was evaluated on two datasets (Kaggle DR 2015 competition (Dataset 1; 34,984 FIs) and APTOS 2019 (3,662 FIs)), and the results are promising. For the two datasets mentioned, the proposed technique attained accuracies of 91.78 % and 97.27 % respectively. However, one of the study's subsidiary discoveries was that the proposed framework demonstrated stability for both larger and smaller datasets, as well as for balanced and imbalanced datasets. Furthermore, in terms of classifier performance metrics, model parameters and layers, and prediction time, the suggested approach outscored existing state-of-the-art models, which would add significant benefit for the medical practitioners in accurately identifying the DR.
A Novel Image Registration Framework for Monitoring Diabetic Retinopathy Progression with Genetic Algorithm-Based Image Alignment
2024, International Journal of Intelligent Systems and Applications in Engineering

View all citing articles on Scopus

View full text

Automated detection of diabetic retinopathy using convolutional neural networks on a small dataset

Highlights

Abstract

Introduction

Section snippets

Related work

Method and structure

Results

Discussion and conclusion

Declaration of Competing Interest

Ophthalmology

Deep learning

Nature

Automated identification of diabetic retinopathy stages using digital fundus images

J. Med. Syst.

Application of higher order spectra for the identification of diabetes retinopathy stages

J. Med. Syst.

Multiclass svm-based automated diagnosis of diabetic retinopathy

Algorithms for the automated detection of diabetic retinopathy using digital fundus images: a review

J. Med. Syst.