Automated diabetic retinopathy screening in resource-limited areas with attention-enhanced deep learning on fundus images DOI

Sornil A. Binusha,

Herobin Rani C. Sheeja,

Sheeba I. Rexiline

et al.

i-manager’s Journal on Image Processing, Journal Year: 2024, Volume and Issue: 11(4), P. 10 - 10

Published: Jan. 1, 2024

Diabetic retinopathy (DR) is a leading contributor to vision impairment, particularly in areas with limited resources where access specialized care scarce. This study introduces an automated screening system for DR using attention- enhanced deep learning on retinal fundus images, specifically designed these regions. The leverages convolutional neural network (CNN) technology integrated attention mechanisms focus critical features indicative of DR, such as microaneurysms and hemorrhages, improving detection accuracy reliability. Varied images were used training validation, data augmentation applied enhance model robustness. was optimized deployment low-cost hardware, ensuring feasibility resource-limited settings. Performance evaluation demonstrated high sensitivity specificity, maps provided interpretability healthcare providers. has the potential early diabetic underserved areas, facilitating timely intervention reducing risk blindness. By making advanced diagnostic tools accessible, this approach promotes equitable helps prevent loss globally.

Language: Английский

Predicting coronary heart disease with advanced machine learning classifiers for improved cardiovascular risk assessment DOI Creative Commons
M. Rehman, Shahid Naseem, Ateeq Ur Rehman Butt

et al.

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: April 17, 2025

Worldwide, coronary heart disease (CHD) is a leading cause of mortality, and its early prediction remains critical challenge in clinical data analysis. Machine learning (ML) offers valuable diagnostic support by leveraging healthcare to enhance decision-making accuracy. Although numerous studies have applied ML classifiers for prediction, their contributions often lack clarity addressing key challenges. In this paper, we present comprehensive framework that systematically tackles these issues. First, employ mutual information (MI) effective feature selection isolate the most informative predictors. Second, address significant class imbalance dataset using Synthetic Minority Oversampling Technique (SMOTE), which substantially improves model training. Third, propose novel hybrid integrates particle swarm optimization (PSO) with an artificial neural network (ANN) optimize weighting bias Additionally, conduct comparative analysis traditional classifiers, including Logistic Regression Random Forest, National Health Nutritional Examination Survey dataset. Our results demonstrate while conventional achieve accuracy 95.8%, proposed PSO-ANN attains enhanced up 97% predicting CHD. This work clearly defines improving selection, handling imbalance, introducing innovative superior performance.

Language: Английский

Citations

0
Enhanced multi-grade diabetic retinopathy detection and classification via ensembled deep learning model from retinal fundus images DOI

Peddapullaiahgari Hariobulesu,

Fahimuddin Shaik

Expert Systems with Applications, Journal Year: 2025, Volume and Issue: unknown, P. 128116 - 128116

Published: May 1, 2025

Language: Английский

Citations

0
Pixel attention meets M-shaped networks: A cutting-edge AI solution for diabetic retinopathy classification and stroke risk prediction DOI
Ahmed A. Alsheikhy, Tawfeeq Shawly, Yahia Said

et al.

Biomedical Signal Processing and Control, Journal Year: 2025, Volume and Issue: 110, P. 108110 - 108110

Published: May 27, 2025

Language: Английский

Citations

0
Artificial Intelligence (AI)-Enhanced Detection of Diabetic Retinopathy From Fundus Images: The Current Landscape and Future Directions DOI Open Access

Lara Alsadoun,

Husnain Ali,

Muhammad Muaz Mushtaq

et al.

Cureus, Journal Year: 2024, Volume and Issue: unknown

Published: Aug. 26, 2024

Diabetic retinopathy (DR) remains a leading cause of vision loss worldwide, with early detection critical for preventing irreversible damage. This review explores the current landscape and future directions artificial intelligence (AI)-enhanced DR from fundus images. Recent advances in deep learning computer have enabled AI systems to analyze retinal images expert-level accuracy, potentially transforming screening. Key developments include convolutional neural networks achieving high sensitivity specificity detecting referable DR, multi-task approaches that can simultaneously detect grade severity, lightweight models enabling deployment on mobile devices. While these show promise improving efficiency accessibility screening, several challenges remain. These ensuring generalizability across diverse populations, standardizing image acquisition quality, addressing "black box" nature complex models, integrating seamlessly into clinical workflows. Future field encompass explainable enhance transparency, federated leverage decentralized datasets, integration electronic health records other diagnostic modalities. There is also growing potential contribute personalized treatment planning predictive analytics disease progression. As technology continues evolve, maintaining focus rigorous validation, ethical considerations, real-world implementation will be crucial realizing full AI-enhanced global eye outcomes.

Language: Английский

Citations

2
Automatic Segmentation and Statistical Analysis of the Foveal Avascular Zone DOI Creative Commons

Geanina Totolici,

Mihaela Miron,

Anisia-Luiza Culea-Florescu

et al.

Technologies, Journal Year: 2024, Volume and Issue: 12(12), P. 235 - 235

Published: Nov. 21, 2024

This study facilitates the extraction of foveal avascular zone (FAZ) metrics from optical coherence tomography angiography (OCTA) images, offering valuable clinical insights and enabling detailed statistical analysis FAZ size shape across three patient groups: healthy, type II diabetes mellitus both (DM) high blood pressure (HBP). Additionally, it evaluates performance four deep learning (DL) models—U-Net, U-Net with DenseNet121, MobileNetV2 VGG16—in automating segmentation FAZ. Manual images by ophthalmological clinicians was performed initially, data augmentation used to enhance dataset for robust model training evaluation. Consequently, original set 103 full retina OCTA extended 672 cases, including 42 normal patients, 357 DM 273 patients HBP. Among models, DenseNet outperformed others, achieving highest accuracy, Intersection over Union (IoU), Dice coefficient all groups. research is distinct in its focus on inclusion hypertension diabetes, an area that less studied existing literature.

Language: Английский

Citations

1
Detection of Diabetic Retinopathy Using Deep Learning DOI Creative Commons
A. Sabo,

Muhammadul Habib Bn Umar,

Swati Sah

et al.

Cureus Journal of Computer Science., Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 11, 2024

Language: Английский

Citations

0
Detection of Diabetic Retinopathy Using Deep Learning DOI Creative Commons
A. Sabo,

Muhammadul Habib Bn Umar,

Swati Sah

et al.

Cureus Journal of Computer Science., Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 11, 2024

Language: Английский

Citations

0
Automated diabetic retinopathy screening in resource-limited areas with attention-enhanced deep learning on fundus images DOI

Sornil A. Binusha,

Herobin Rani C. Sheeja,

Sheeba I. Rexiline

et al.

i-manager’s Journal on Image Processing, Journal Year: 2024, Volume and Issue: 11(4), P. 10 - 10

Published: Jan. 1, 2024

Diabetic retinopathy (DR) is a leading contributor to vision impairment, particularly in areas with limited resources where access specialized care scarce. This study introduces an automated screening system for DR using attention- enhanced deep learning on retinal fundus images, specifically designed these regions. The leverages convolutional neural network (CNN) technology integrated attention mechanisms focus critical features indicative of DR, such as microaneurysms and hemorrhages, improving detection accuracy reliability. Varied images were used training validation, data augmentation applied enhance model robustness. was optimized deployment low-cost hardware, ensuring feasibility resource-limited settings. Performance evaluation demonstrated high sensitivity specificity, maps provided interpretability healthcare providers. has the potential early diabetic underserved areas, facilitating timely intervention reducing risk blindness. By making advanced diagnostic tools accessible, this approach promotes equitable helps prevent loss globally.

Language: Английский

Citations

0