Enhancing Early Detection of Diabetic Retinopathy Through the Integration of Deep Learning Models and Explainable Artificial Intelligence
Artículo
Materias > Ingeniería
Universidad Europea del Atlántico > Investigación > Producción Científica
Universidad Internacional Iberoamericana México > Investigación > Producción Científica
Universidad de La Romana > Investigación > Producción Científica
Abierto
Inglés
Humans can carry various diseases, some of which are poorly understood and lack comprehensive solutions. Such a disease can exists in human eye that can affect one or both eyes is diabetic retinopathy (DR) which can impair function, vision, and eventually result in permanent blindness. It is one of those complex complexities. Therefore, early detection of DR can significantly reduce the risk of vision impairment by appropriate treatment and necessary precautions. The primary aim of this study is to leverage cutting-edge models trained on diverse image datasets and propose a CNN model that demonstrates comparable performance. Specifically, we employ transfer learning models such as DenseNet121, Xception, Resnet50, VGG16, VGG19, and InceptionV3, and machine learning models such as SVM, and neural network models like (RNN) for binary and multi-class classification. It has been shown that the proposed approach of multi-label classification with softmax functions and categorical cross-entropy works more effectively, yielding perfect accuracy, precision, and recall values. In particular, Xception achieved an impressive 82% accuracy among all the transfer learning models, setting a new benchmark for the dataset used. However, our proposed CNN model shows superior performance, achieving an accuracy of 95.27% on this dataset, surpassing the state-of-the-art Xception model. Moreover, for single-label (binary classifications), our proposed model achieved perfect accuracy as well. Through exploration of these advances, our objective is to provide a comprehensive overview of the leading methods for the early detection of DR. The aim is to discuss the challenges associated with these methods and highlight potential enhancements. In essence, this paper provides a high-level perspective on the integration of deep learning techniques and machine learning models, coupled with explainable artificial intelligence (XAI) and gradient-weighted class activation mapping (Grad-CAM). We prese...
metadata
Ahnaf Alavee, Kazi; Hasan, Mehedi; Hasnayen Zillanee, Abu; Mostakim, Moin; Uddin, Jia; Silva Alvarado, Eduardo René; de la Torre Diez, Isabel; Ashraf, Imran y Abdus Samad, Md
mail
SIN ESPECIFICAR, SIN ESPECIFICAR, SIN ESPECIFICAR, SIN ESPECIFICAR, SIN ESPECIFICAR, eduardo.silva@funiber.org, SIN ESPECIFICAR, SIN ESPECIFICAR, SIN ESPECIFICAR
(2024)
Enhancing Early Detection of Diabetic Retinopathy Through the Integration of Deep Learning Models and Explainable Artificial Intelligence.
IEEE Access, 12.
pp. 73950-73969.
ISSN 2169-3536
|
Texto
s41598-024-63831-0 (1).pdf Available under License Creative Commons Attribution Non-commercial No Derivatives. Descargar (6MB) | Vista Previa |
Resumen
Humans can carry various diseases, some of which are poorly understood and lack comprehensive solutions. Such a disease can exists in human eye that can affect one or both eyes is diabetic retinopathy (DR) which can impair function, vision, and eventually result in permanent blindness. It is one of those complex complexities. Therefore, early detection of DR can significantly reduce the risk of vision impairment by appropriate treatment and necessary precautions. The primary aim of this study is to leverage cutting-edge models trained on diverse image datasets and propose a CNN model that demonstrates comparable performance. Specifically, we employ transfer learning models such as DenseNet121, Xception, Resnet50, VGG16, VGG19, and InceptionV3, and machine learning models such as SVM, and neural network models like (RNN) for binary and multi-class classification. It has been shown that the proposed approach of multi-label classification with softmax functions and categorical cross-entropy works more effectively, yielding perfect accuracy, precision, and recall values. In particular, Xception achieved an impressive 82% accuracy among all the transfer learning models, setting a new benchmark for the dataset used. However, our proposed CNN model shows superior performance, achieving an accuracy of 95.27% on this dataset, surpassing the state-of-the-art Xception model. Moreover, for single-label (binary classifications), our proposed model achieved perfect accuracy as well. Through exploration of these advances, our objective is to provide a comprehensive overview of the leading methods for the early detection of DR. The aim is to discuss the challenges associated with these methods and highlight potential enhancements. In essence, this paper provides a high-level perspective on the integration of deep learning techniques and machine learning models, coupled with explainable artificial intelligence (XAI) and gradient-weighted class activation mapping (Grad-CAM). We prese...
| Tipo de Documento: | Artículo |
|---|---|
| Palabras Clave: | Diabetic retinopathy, transfer learning, CNN, Xception, inception, Grad-CAM |
| Clasificación temática: | Materias > Ingeniería |
| Divisiones: | Universidad Europea del Atlántico > Investigación > Producción Científica Universidad Internacional Iberoamericana México > Investigación > Producción Científica Universidad de La Romana > Investigación > Producción Científica |
| Depositado: | 17 Jun 2024 23:30 |
| Ultima Modificación: | 17 Jun 2024 23:30 |
| URI: | https://repositorio.uniromana.edu.do/id/eprint/12752 |
Acciones (logins necesarios)
 |
Ver Objeto |
en
close
Single-cell omics for nutrition research: an emerging opportunity for human-centric investigations
Understanding how dietary compounds affect human health is challenged by their molecular complexity and cell-type–specific effects. Conventional multi-cell type (bulk) analyses obscure cellular heterogeneity, while animal and standard in vitro models often fail to replicate human physiology. Single-cell omics technologies—such as single-cell RNA sequencing, as well as single-cell–resolved proteomic and metabolomic approaches—enable high-resolution investigation of nutrient–cell interactions and reveal mechanisms at a single-cell resolution. When combined with advanced human-derived in vitro systems like organoids and organ-on-chip platforms, they support mechanistic studies in physiologically relevant contexts. This review outlines emerging applications of single-cell omics in nutrition research, emphasizing their potential to uncover cell-specific dietary responses, identify nutrient-sensitive pathways, and capture interindividual variability. It also discusses key challenges—including technical limitations, model selection, and institutional biases—and identifies strategic directions to facilitate broader adoption in the field. Collectively, single-cell omics offer a transformative framework to advance human-centric nutrition research.
Manuela Cassotta mail manucassotta@gmail.com, Yasmany Armas Diaz mail , Danila Cianciosi mail , Bei Yang mail , Zexiu Qi mail , Ge Chen mail , Santos Gracia Villar mail santos.gracia@uneatlantico.es, Luis Alonso Dzul López mail luis.dzul@uneatlantico.es, Giuseppe Grosso mail , José L. Quiles mail , Jianbo Xiao mail , Maurizio Battino mail maurizio.battino@uneatlantico.es, Francesca Giampieri mail francesca.giampieri@uneatlantico.es,
Cassotta
<a href="/17862/1/sensors-25-06419.pdf" class="ep_document_link"><img class="ep_doc_icon" alt="[img]" src="/17862/1.hassmallThumbnailVersion/sensors-25-06419.pdf" border="0"/></a>
en
open
Edge-Based Autonomous Fire and Smoke Detection Using MobileNetV2
Forest fires pose significant threats to ecosystems, human life, and the global climate, necessitating rapid and reliable detection systems. Traditional fire detection approaches, including sensor networks, satellite monitoring, and centralized image analysis, often suffer from delayed response, high false positives, and limited deployment in remote areas. Recent deep learning-based methods offer high classification accuracy but are typically computationally intensive and unsuitable for low-power, real-time edge devices. This study presents an autonomous, edge-based forest fire and smoke detection system using a lightweight MobileNetV2 convolutional neural network. The model is trained on a balanced dataset of fire, smoke, and non-fire images and optimized for deployment on resource-constrained edge devices. The system performs near real-time inference, achieving a test accuracy of 97.98% with an average end-to-end prediction latency of 0.77 s per frame (approximately 1.3 FPS) on the Raspberry Pi 5 edge device. Predictions include the class label, confidence score, and timestamp, all generated locally without reliance on cloud connectivity, thereby enhancing security and robustness against potential cyber threats. Experimental results demonstrate that the proposed solution maintains high predictive performance comparable to state-of-the-art methods while providing efficient, offline operation suitable for real-world environmental monitoring and early wildfire mitigation. This approach enables cost-effective, scalable deployment in remote forest regions, combining accuracy, speed, and autonomous edge processing for timely fire and smoke detection.
Dilshod Sharobiddinov mail , Hafeez Ur Rehman Siddiqui mail , Adil Ali Saleem mail , Gerardo Méndez Mezquita mail , Debora L. Ramírez-Vargas mail debora.ramirez@unini.edu.mx, Isabel de la Torre Díez mail ,
Sharobiddinov
<a href="/17863/1/v16p4316.pdf" class="ep_document_link"><img class="ep_doc_icon" alt="[img]" src="/17863/1.hassmallThumbnailVersion/v16p4316.pdf" border="0"/></a>
en
open
Breast cancer is a lethal carcinoma impacting a considerable number of women across the globe. While preventive measures are limited, early detection remains the most effective strategy. Accurate classification of breast tumors into benign and malignant categories is important which may help physicians in diagnosing the disease faster. This survey investigates the emerging inclination and approaches in the area of machine learning (ML) for the diagnosis of breast cancer, pointing out the classification techniques based on both segmentation and feature selection. Certain datasets such as the Wisconsin Diagnostic Breast Cancer Dataset (WDBC), Wisconsin Breast Cancer Dataset Original (WBCD), Wisconsin Prognostic Breast Cancer Dataset (WPBC), BreakHis, and others are being evaluated in this study for the demonstration of their influence on the performance of the diagnostic tools and the accuracy of the models such as Support vector machine, Convolutional Neural Networks (CNNs) and ensemble approaches. The main shortcomings or research gaps such as prejudice of datasets, scarcity of generalizability, and interpretation challenges are highlighted. This research emphasizes the importance of the hybrid methodologies, cross-dataset validation, and the engineering of explainable AI to narrow these gaps and enhance the overall clinical acceptance of ML-based detection tools.
Alveena Saleem mail , Muhammad Umair mail , Muhammad Tahir Naseem mail , Muhammad Zubair mail , Silvia Aparicio Obregón mail silvia.aparicio@uneatlantico.es, Rubén Calderón Iglesias mail ruben.calderon@uneatlantico.es, Shoaib Hassan mail , Imran Ashraf mail ,
Saleem
<a class="ep_document_link" href="/17849/1/1-s2.0-S2590005625001043-main.pdf"><img class="ep_doc_icon" alt="[img]" src="/17849/1.hassmallThumbnailVersion/1-s2.0-S2590005625001043-main.pdf" border="0"/></a>
en
open
Ultra Wideband radar-based gait analysis for gender classification using artificial intelligence
Gender classification plays a vital role in various applications, particularly in security and healthcare. While several biometric methods such as facial recognition, voice analysis, activity monitoring, and gait recognition are commonly used, their accuracy and reliability often suffer due to challenges like body part occlusion, high computational costs, and recognition errors. This study investigates gender classification using gait data captured by Ultra-Wideband radar, offering a non-intrusive and occlusion-resilient alternative to traditional biometric methods. A dataset comprising 163 participants was collected, and the radar signals underwent preprocessing, including clutter suppression and peak detection, to isolate meaningful gait cycles. Spectral features extracted from these cycles were transformed using a novel integration of Feedforward Artificial Neural Networks and Random Forests , enhancing discriminative power. Among the models evaluated, the Random Forest classifier demonstrated superior performance, achieving 94.68% accuracy and a cross-validation score of 0.93. The study highlights the effectiveness of Ultra-wideband radar and the proposed transformation framework in advancing robust gender classification.
Adil Ali Saleem mail , Hafeez Ur Rehman Siddiqui mail , Muhammad Amjad Raza mail , Sandra Dudley mail , Julio César Martínez Espinosa mail ulio.martinez@unini.edu.mx, Luis Alonso Dzul López mail luis.dzul@uneatlantico.es, Isabel de la Torre Díez mail ,
Saleem
<a class="ep_document_link" href="/17857/1/excli2025-8779.pdf"><img class="ep_doc_icon" alt="[img]" src="/17857/1.hassmallThumbnailVersion/excli2025-8779.pdf" border="0"/></a>
en
open
Polyphenols are naturally occurring compounds that can be found in plant-based foods, including fruits, vegetables, nuts, seeds, herbs, spices, and beverages, the use of which has been linked to enhanced brain health and cognitive function. These natural molecules are broadly classified into two main groups: flavonoids and non-flavonoid polyphenols, the latter including phenolic acids, stilbenes, and tannins. Flavonoids are primarily known for their potent antioxidant properties, which help neutralize harmful reactive oxygen species (ROS) in the brain, thereby reducing oxidative stress, a key contributor to neurodegenerative diseases. In addition to their antioxidant effects, flavonoids have been shown to modulate inflammation, enhance neuronal survival, and support neurogenesis, all of which are critical for maintaining cognitive function. Phenolic acids possess strong antioxidant properties and are believed to protect brain cells from oxidative damage. Neuroprotective effects of these molecules can also depend on their ability to modulate signaling pathways associated with inflammation and neuronal apoptosis. Among polyphenols, hydroxycinnamic acids such as caffeic acid have been shown to enhance blood-brain barrier permeability, which may increase the delivery of other protective compounds to the brain. Another compound of interest is represented by resveratrol, a stilbene extensively studied for its potential neuroprotective properties related to its ability to activate the sirtuin pathway, a molecular signaling pathway involved in cellular stress response and aging. Lignans, on the other hand, have shown promise in reducing neuroinflammation and oxidative stress, which could help slow the progression of neurodegenerative diseases and cognitive decline. Polyphenols belonging to different subclasses, such as flavonoids, phenolic acids, stilbenes, and lignans, exert neuroprotective effects by regulating microglial activation, suppressing pro-inflammatory cytokines, and mitigating oxidative stress. These compounds act through multiple signaling pathways, including NF-κB, MAPK, and Nrf2, and they may also influence genetic regulation of inflammation and immune responses at brain level. Despite their potential for brain health and cognitive function, polyphenols are often characterized by low bioavailability, something that deserves attention when considering their therapeutic potential. Future translational studies are needed to better understand the right dosage, the overall diet, the correct target population, as well as ideal formulations allowing to overcome bioavailability limitations.
Justyna Godos mail , Giuseppe Carota mail , Giuseppe Caruso mail , Agnieszka Micek mail , Evelyn Frias-Toral mail , Francesca Giampieri mail francesca.giampieri@uneatlantico.es, Julién Brito Ballester mail julien.brito@uneatlantico.es, Maurizio Battino mail maurizio.battino@uneatlantico.es, Carmen Lilí Rodríguez Velasco mail carmen.rodriguez@uneatlantico.es, José L. Quiles mail jose.quiles@uneatlantico.es,
Godos