Mohamed TOUATI

Dr. Mohamed TOUATI

University of Western Brittany, Brest, Bretagne, France
orcid

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
Verified
Journal Article
Open access
PDF file
Mohamed TOUATIMohamed TOUATI
Mohamed TOUATI
RTRabeb Touati
Rabeb Touati
LNLaurent Nana
Laurent Nana
FBFaouzi Benzarti
Faouzi Benzarti
Sadok BEN YAHIASadok BEN YAHIA
Sadok BEN YAHIA
Diabetic retinopathy, a common complication of diabetes, is further exacerbated by factors such as hypertension and obesity. This study introduces the Diabetic Retinopathy Compact Convolutional Transformer (DRCCT) model, which combines convolutional and transformer techniques to enhance the classification of retinal images. The DRCCT model achieved an impressive average F1-score of 0.97, reflecting its high accuracy in detecting true positives while minimizing false positives. Over 100 training epochs, the model demonstrated outstanding generalization capabilities, achieving a remarkable training accuracy of 99% and a validation accuracy of 95%. This consistent improvement underscores the model’s robust learning process and its effectiveness in avoiding overfitting. On a newly evaluated dataset, the model attained precision and recall scores of 96.93% and 98.89%, respectively, indicating a well-balanced handling of false positives and false negatives. The model’s ability to classify retinal images into five distinct diabetic retinopathy categories demonstrates its potential to significantly improve automated diagnosis and aid in clinical decision-making.

28 April 2023
Verified
Book Chapter
Closed access
PDF file
Mohamed TOUATIMohamed TOUATI
Mohamed TOUATI
LNLaurent Nana
Laurent Nana
FBFaouzi Benzarti
Faouzi Benzarti

1 July 2024
Verified
Conference Paper
Closed access
Mohamed TOUATIMohamed TOUATI
Mohamed TOUATI
LNLaurent Nana
Laurent Nana
FBFaouzi Benzarti
Faouzi Benzarti
Diabetic Retinopathy (DR) is a critical health concern that, if diagnosed early, can significantly reduce the risk of blindness among individuals with diabetes. Leveraging the power of artificial intelligence, this paper introduces an innovative approach for DR detection using Deep Learning techniques. Our methodology employs Transfer Learning, adapting a modified ResNet50 model integrated with a self-attention mechanism. This novel strategy not only enhances detection efficiency but also reduces associated costs. The modified model is specifically tailored to address challenges stemming from unbalanced datasets, such as the APTOS 2019 Blindness Detection dataset. By incorporating a self-attention mechanism, the model can better focus on pertinent features, thereby improving classification and detection accuracy. Performance metrics demonstrate the efficacy of our approach, with a training accuracy of 98.24%, a test accuracy of 0.89, and a F1 Score of 0.94.

9 April 2026
Verified
Conference Paper
Closed access
Mohamed TOUATIMohamed TOUATI
Mohamed TOUATI
RTRabeb Touati
Rabeb Touati
MMMohamed Medimegh
Mohamed Medimegh
LNLaurent Nana
Laurent Nana
MMMohamed Wiem Mkaouer
Mohamed Wiem Mkaouer
Working with Pixemantic and its Pixheart platform, this project introduces an AI-driven web app for automated Electrocardiogram (ECG) analysis. The system processes scanned ECG images using CNN and LSTM models to extract cardiac features to detect abnormalities with accuracy levels that match those achieved by medical professionals. Our hybrid CNN-LSTM model achieved an accuracy of 86%, with precision at 88%, recall at 87%, and an F1-score of 87%. Support Vector Machines (SVM) and Logistic Regression achieved strong results in traditional machine learning approaches because SVM models achieved 96% accuracy and Logistic Regression achieved 93%. Random Forest and XGBoost provide distinct advantages: Random Forest achieved perfect MI precision of 1.00 and achieved high PM precision of 0.97 but XGBoost delivered better HB recall with 0.81 and achieved an F1-score of 0.86 . The platform supports real-time analysis, clear waveform visuals, and report generation. Medical specialists who work in clinical environments have verified the data quality together with the model's operational success which proves the system's ability to deliver quick and precise ECG interpretation .

23 June 2022 in Research Square
Verified
Preprint
Open access
PDF file
RTRabeb Touati
Rabeb Touati
Mohamed TOUATIMohamed TOUATI
Mohamed TOUATI
FBFaouzi Benzarti
Faouzi Benzarti
Vishal KumarVishal Kumar
Vishal Kumar
MKMaher Kharrat
Maher Kharrat
Ahmed A. ElngarAhmed A. Elngar
Ahmed A. Elngar
Coronaviruses are a type of frequent RNA virus. They are responsible for digestive and respiratory infections in animal and human genomes. A coronavirus renamed COVID-19 appeared and spread in the world which makes it declared in March 2020 a pandemic by the World Health Organization. SARS-CoV-2 genome, responsible for COVID-19 virus, has a size equal to 29,903 nucleotides, and its genetic makeup is composed of 11 functional open reading frames (ORFs). This paper proposes comparison algorithms to find SARS-Cov2 origin using digital genomic signatures. As a result, the five closest genomes related to SARS-CoV-2 are RaTG13, Pangolin-cov-PCoV_GX-P2V, Pangolin CoV MP789, bat-SL-CoVZC45 and bat-SL-CoVZXC21.

25 October 2025
Verified
Conference Paper
Closed access
Mohamed TOUATIMohamed TOUATI
Mohamed TOUATI
RTRabeb Touati
Rabeb Touati
LNLaurent Nana
Laurent Nana
FBFaouzi Benzarti
Faouzi Benzarti
MMMohamed Wiem Mkaouer
Mohamed Wiem Mkaouer
SYSadok Ben Yahia
Sadok Ben Yahia
Diabetic retinopathy is a leading cause of blindness, it's largely preventable with early detection. Catching it in time is critical to stopping permanent vision loss. Right now, diagnosing it depends on specialists manually examining retinal images. This approach isn't just time-consuming; it's also highly subjective, leaving room for human error and inconsistency. To address these challenges, we propose ARNet-DR, a ResNet50-based architecture enhanced with a kernel attention mechanism for robust $D R$ classification. The attention modules enable effective modeling of contextual dependencies in fundus images, prioritizing subtle yet discriminative features relevant to $D R$ staging. ARNet-DR was assessed on the class-imbalanced APTOS 2019 Blindness Detection dataset, where it attained a training accuracy of 98.24 %, a test accuracy of 94.5%, and an F1-score of 0.99. These outcomes highlight the model's robustness in addressing class imbalance and image diversity, surpassing existing baseline approaches. Owing to its high precision and strong generalization performance, ARNetDR emerges as a viable decision-support aid for ophthalmologists and a significant advancement toward scalable, AI-powered diabetic retinopathy screening.