Deep Learning in Retinal Image Segmentation and Feature Extraction: A Review

Artificial intelligence for diabetic retinopathy screening in Africa

Retinal fundus images play a crucial role in the early detection of eye problems, aiding in timely diagnosis and treatment to prevent vision loss or blindness. With advancements in technology, Convolutional Neural Network (CNN) algorithms have emerged as effective tools for recognition, delineation, and classification tasks. This study proposes a comprehensive review of CNN algorithms used for retinal fundus image segmentation and classification. Our review follows a systematic approach, exploring diverse repositories to identify studies employing CNN to segment and classify retinal fundus images. Utilizing CNNs in the segmentation and classification of retinal fundus images can enhance the precision of segmentation outcomes and alleviate the sole dependence on human experts. This approach enables more accurate segmentation results, reducing the burden on human experts. A total of sixty-two studies are included in our review, analyzing aspects such as database usage and the advantages and disadvantages of the methods employed. The review provides valuable insights, limitations, observations, and future directions in the field. Despite certain limitations, the findings indicate that CNN algorithms consistently achieve high accuracies. The comprehensive examination of the included studies sheds light on the potential of CNN in retinal fundus image analysis.

The analysis of retinal vessel images plays a crucial role in the early diagnosis of various diseases. Existing retinal vascular segmentation algorithms rely only on convolutional neural networks (CNN) to carry out feature extraction. However, due to the local nature of the convolutional operation, the CNN-based algorithms are difficult to extract the global features of the picture and are insensitive to the position information of each feature, which leads to poor connectivity among the segmented retinal vessels and insufficient segmentation of fine vessels. Therefore, we propose a retinal vessel segmentation algorithm based on an axial transformer and CNN. In this algorithm, the axial transformer is used to encode the position of each feature of the image and extract their global features, and the CNN is used to extract the local features of the image. The experimental results on the CHASE DB1 dataset indicate that the proposed algorithm can effectively segment the fine vessels in retinal images and significantly improve the segmentation accuracy of retinal vessels.

Using Deep Learning Models to Characterize Major Retinal Features on Color Fundus Photographs

Accurate segmentation of retinal vessels is critical to the mechanism, diagnosis, and treatment of many ocular pathologies. Due to the poor contrast and inhomogeneous background of fundus imaging and the complex structure of retinal fundus images, this makes accurate segmentation of blood vessels from retinal images still challenging. In this paper, we propose an effective framework for retinal vascular segmentation, which is innovative mainly in the retinal image pre-processing stage and segmentation stage. First, we perform image enhancement on three publicly available fundus datasets based on the multiscale retinex with color restoration (MSRCR) method, which effectively suppresses noise and highlights the vessel structure creating a good basis for the segmentation phase. The processed fundus images are then fed into an effective Reverse Fusion Attention Residual Network (RFARN) for training to achieve more accurate retinal vessel segmentation. In the RFARN, we use Reverse Channel Attention Module (RCAM) and Reverse Spatial Attention Module (RSAM) to highlight the shallow details of the channel and spatial dimensions. And RCAM and RSAM are used to achieve effective fusion of deep local features with shallow global features to ensure the continuity and integrity of the segmented vessels. In the experimental results for the DRIVE, STARE and CHASE datasets, the evaluation metrics were 0.9712, 0.9822 and 0.9780 for accuracy (Acc), 0.8788, 0.8874 and 0.8352 for sensitivity (Se), 0.9803, 0.9891 and 0.9890 for specificity (Sp), area under the ROC curve(AUC) was 0.9910, 0.9952 and 0.9904, and the F1-Score was 0.8453, 0.8707 and 0.8185. In comparison with existing retinal image segmentation methods, e.g. UNet, R2UNet, DUNet, HAnet, Sine-Net, FANet, etc., our method in three fundus datasets achieved better vessel segmentation performance and results.

This paper presents Deep Retinal Image Segmentation, a unified framework of retinal image analysis that provides both optic disc and exudates segmentation. The paper presents a new formulation of fully Convolutional Neural Networks (FCNs) that allows accurate segmentation of the retinal images. A major modification in these retinal image segmentation tasks are to improve and speed-up the FCNs training by adding short and long skip connections in standard FCNs architecture with class-balancing loss. The proposed method is experimented on the DRIONS-DB dataset for optic disc segmentation and the privately dataset for exudates segmentation, which achieves strong performance and significantly outperforms the-state-of-the-art. It achieves 93.12% sensitivity (Sen), 99.56% specificity (Spe), 89.90% Positive predictive value (PPV) and 90.93% F-score for optic disc segmentation while 81.35% Sen, 98.76% Spe, 81.64% PPV and 81.50% F-score for exudates segmentation respectively.

PurposeTo propose a novel deep learning–based methodology for drusen detection and quantification in early age-related macular degeneration (AMD) using retinal multispectral images. The retinal multispectral images highlight features in several nonoverlapping spectral bands that the deep learning models leverage for automatic drusen detection and quantification in dry AMD.MethodsThe proposed novel methodology comprises quality assessment of retinal images, region of interest extraction, drusen segmentation, and drusen quantification stages. Different deep learning models (such as UNet++ convolutional neural network with EfficientNetV2 encoder) have been implemented for these stages. A total of 170 drusen and 150 nondrusen retinal images (single eye) were split into four training and validation data sets to analyze the performance of a deep learning model for drusen segmentation.ResultsThe proposed methodology achieved an average score, recall, and precision of 0.691, 0.668, and 0.776, respectively, across all four validation sets. This work also analyzed the performance of the proposed deep learning model for discriminating drusen and drusen-like lesions, achieving a pixel-wise segmentation accuracy of 99.998%. The number and the diameter of the detected drusen were also computed. A Dice score distribution for drusen segmentation with different numbers and sizes of drusen per eye is also shown.ConclusionsThis work demonstrates that deep learning models applied to retinal multispectral images can provide accurate and clinically significant drusen segmentation and quantification, thereby facilitating early detection, longitudinal monitoring, and reduction of the risk of vision loss from AMD.Translational RelevanceDeep learning–assisted detection of drusen from multispectral retinal images will refine and improve clinical diagnosis of early nonexudative age-related macular degeneration.

- This paper presents a comparative analysis between a traditional convolutional neural network (CNN) model and a hybrid ResNet50 + AdaBoost architecture for non-invasive cardiovascular risk prediction using retinal fundus images. The baseline model, trained on a small dataset for binary classification, achieves reasonable accuracy but lacks scalability and granularity. In contrast, the proposed model leverages deep residual learning for feature extraction and ensemble-based classification to predict five distinct cardiovascular risk levels. Evaluated on a class-balanced dataset of over 11,500 images, the hybrid model achieved a testing accuracy of 91.48% and a macro F1-score of approximately 0.93. The system also integrates data visualization and database logging to support clinical explainability. The findings demonstrate the proposed method’s superiority in risk stratification, robustness, and deployment readiness for real-world applications. Key Words: Retinal Imaging, Cardiovascular Risk Prediction, ResNet50, AdaBoost, Deep Learning, Ensemble Learning, Fundus Images, Heart Attack Classification

Background and Objectives: Brain Tumor Fusion-based Segments and Classification-Non-enhancing tumor (BTFSC-Net) is a hybrid system for classifying brain tumors that combine medical image fusion, segmentation, feature extraction, and classification procedures. Materials and Methods: to reduce noise from medical images, the hybrid probabilistic wiener filter (HPWF) is first applied as a preprocessing step. Then, to combine robust edge analysis (REA) properties in magnetic resonance imaging (MRI) and computed tomography (CT) medical images, a fusion network based on deep learning convolutional neural networks (DLCNN) is developed. Here, the brain images’ slopes and borders are detected using REA. To separate the sick region from the color image, adaptive fuzzy c-means integrated k-means (HFCMIK) clustering is then implemented. To extract hybrid features from the fused image, low-level features based on the redundant discrete wavelet transform (RDWT), empirical color features, and texture characteristics based on the gray-level cooccurrence matrix (GLCM) are also used. Finally, to distinguish between benign and malignant tumors, a deep learning probabilistic neural network (DLPNN) is deployed. Results: according to the findings, the suggested BTFSC-Net model performed better than more traditional preprocessing, fusion, segmentation, and classification techniques. Additionally, 99.21% segmentation accuracy and 99.46% classification accuracy were reached using the proposed BTFSC-Net model. Conclusions: earlier approaches have not performed as well as our presented method for image fusion, segmentation, feature extraction, classification operations, and brain tumor classification. These results illustrate that the designed approach performed more effectively in terms of enhanced quantitative evaluation with better accuracy as well as visual performance.

Artificial intelligence using deep learning to screen for referable and vision-threatening diabetic retinopathy in Africa: a clinical validation study

Identification of retinal diseases is a test for the ophthalmologists as the anomalies are just unmistakable at the beginning period. Early detection of these diseases can avoid lasting vision misfortune. Dealing with a lot of retinal images and location of variations from the norm because of these infections is difficult just as tedious. In this work, the deep learning algorithm has proposed to check the abnormality condition of retina with the help of retinal fundus images. In deep leaning a training set is produced with features of variations from the norm present in the retinal images and the infection the retina is experiencing. The deep Convolutional Neural Network (CNN) classifier predicts the infection for every retinal images in the wake of social event the learning from training the set. The rightness of desire is resolved to evaluate the viability of the classifier. The proposed technique was executed in MATLAB and assessed both normal and abnormal diabetic retinopathy retinal images of IDRID, ROC, and local datasets. The proposed technique has gotten better execution measurements, for example, sensitivity of 98.2%, Specificity of 98.45%, accuracy of 98.56% and average area under receiver operating characteristics of 0.9 when contrasted with different conditions of the workmanship strategies.

Early and accurate identification of brain tumors is essential for determining effective treatment strategies and improving patient outcomes. Artificial intelligence (AI) and deep learning (DL) techniques have shown promise in automating diagnostic tasks based on magnetic resonance imaging (MRI). This study evaluates the performance of four pre-trained deep convolutional neural network (CNN) architectures for the automatic multi-class classification of brain tumors into four categories: Glioma, Meningioma, Pituitary, and No Tumor. The proposed approach utilizes the publicly accessible Brain Tumor MRI Msoud dataset, consisting of 7023 images, with 5712 provided for training and 1311 for testing. To assess the impact of data availability, subsets containing 25%, 50%, 75%, and 100% of the training data were used. A stratified five-fold cross-validation technique was applied. The CNN architectures evaluated include DeiT3_base_patch16_224, Xception41, Inception_v4, and Swin_Tiny_Patch4_Window7_224, all fine-tuned using transfer learning. The training pipeline incorporated advanced preprocessing and image data augmentation techniques to enhance robustness and mitigate overfitting. Among the models tested, Swin_Tiny_Patch4_Window7_224 achieved the highest classification Accuracy of 99.24% on the test set using 75% of the training data. This model demonstrated superior generalization across all tumor classes and effectively addressed class imbalance issues. Furthermore, we deployed and benchmarked the best-performing DL model on embedded AI platforms (Jetson AGX Xavier and Orin Nano), demonstrating their capability for real-time inference and highlighting their feasibility for edge-based clinical deployment. The results highlight the strong potential of pre-trained deep CNN and transformer-based architectures in medical image analysis. The proposed approach provides a scalable and energy-efficient solution for automated brain tumor diagnosis, facilitating the integration of AI into clinical workflows.

In this paper, we look at how to design and build a system to find tumors using 2 Convolutional Neural Network (CNN) models. With the help of digital image processing and deep Learning, we can make a system that automatically diagnoses and finds different diseases and abnormalities. The tumor detection system may include image enhancement, segmentation, data enhancement, feature extraction, and classification. These options are set up so that the CNN model can give the best results. During the training phase, the learning rate is used to change the weights and bias. The learning rate also changes the weights. One Epoch is when all of the training images are shown to the model. As the training data may be very large, the data in each epoch are split into batches. Every epoch has a training session and a test session. After each epoch, the weights are changed based on how fast the CNN is learning. This is done with the help of optimization algorithms. The suggested technique uses the anticipated mean intersection over union value to identify failure instances in addition to forecasting the mean intersection over union. This paper talks about how to separate brain tumors from magnetic resonance images of patients taken from "Brain web." Using basic ideas of digital image processing, magnetic resonance images are used to extract and find tumors using a hybrid method. In this paper, the proposed algorithm is applied with the help of MATLAB. In medical image processing, brain tumor segmentation is an important task. The goal of this paper is to look at different ways to divide brain tumors using magnetic resonance imaging. Recently, automatic segmentation using deep learning methods has become popular because these methods get the best results and are better at solving this problem than others. Deep learning methods can also be used to process and evaluate large amounts of magnetic resonance imaging image data quickly and objectively. A classification method based on a convolution neural network is also added to the proposed scheme to make it more accurate and cut down on the amount of time it takes to do the calculations. Also, the results of the classification are given as images of a tumor or a healthy brain. The training is 98.5% correct. In the same way, both the validation accuracy and validation loss are high.

Random color transformation for single domain generalized retinal image segmentation

Exudates are a common sign of diabetic retinopathy, which is a disease that affects the blood vessels in the retina. Early detection of exudates is critical to avoiding vision problems through continuous screening and treatment. In traditional clinical practice, the involved lesions are manually detected using photographs of the fundus. However, this task is cumbersome and time-consuming and requires intense effort due to the small size of the lesion and the low contrast of the images. Thus, computer-assisted diagnosis of retinal disease based on the detection of red lesions has been actively explored recently. In this paper, we present a comparison of deep convolutional neural network (CNN) architectures and propose a residual CNN with residual skip connections to reduce the parameter for the semantic segmentation of exudates in retinal images. A suitable image augmentation technique is used to improve the performance of network architecture. The proposed network can robustly segment exudates with high accuracy, which makes it suitable for diabetic retinopathy screening. A comparative performance analysis of three benchmark databases: E-ophtha, DIARETDB1, and Hamilton Ophthalmology Institute's Macular Edema, is presented. The proposed method achieves a precision of 0.95, 0.92, 0.97, accuracy of 0.98, 0.98, 0.98, sensitivity of 0.97, 0.95, 0.95, specificity of 0.99, 0.99, 0.99, and area under the curve of 0.97, 0.94, and 0.96, respectively. RESEARCH HIGHLIGHTS: The research focuses on the detection and segmentation of exudates in diabetic retinopathy, a disease affecting the retina. Early detection of exudates is important to avoid vision problems and requires continuous screening and treatment. Currently, manual detection is time-consuming and requires intense effort. The authors compare qualitative results of the state-of-the-art convolutional neural network (CNN) architectures and propose a computer-assisted diagnosis approach based on deep learning, using a residual CNN with residual skip connections to reduce parameters. The proposed method is evaluated on three benchmark databases and demonstrates high accuracy and suitability for diabetic retinopathy screening.