Salwa Said

Dr. Salwa Said

orcid

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

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

Representative Publications

View All Publications
21 January 2026 in Biomedicines
Verified
Journal Article
Open access
PDF file
WAWiem Abdelbaki
Wiem Abdelbaki
HAHend Alshaya
Hend Alshaya
Inzamam Mashood NasirInzamam Mashood Nasir
Inzamam Mashood Nasir
STSara Tehsin
Sara Tehsin
SSSalwa Said
Salwa Said
WBWided Bouchelligua
Wided Bouchelligua
Background: Diagnosing CNS tumors through MRI is limited by significant variability in scanner hardware, acquisition protocols, and intensity characteristics at clinical centers, resulting in substantial domain shifts that lead to diminished reliability for automated models. Methods: We present a Domain-Adaptive MRI Learning Model (DA-MLM) consisting of an adversarially aligned hybrid 3D CNN–transformer encoder with contrastive regularization and covariance-based feature harmonization. Varying sequence MRI inputs (T1, T1ce, T2, and FLAIR) were inputted to multi-scale convolutional layers followed by global self-attention to effectively capture localized tumor structure and long-range spatial context, with domain adaptation that harmonizes feature distribution across datasets. Results: On the BraTS 2020 dataset, we found DA-MLM achieved 94.8% accuracy, 93.6% macro-F1, and 96.2% AUC, improving upon previously established benchmarks by 2–4%. DA-MLM also attained Dice score segmentation of 93.1% (WT), 91.4% (TC), and 89.5% (ET), improving upon 2–3.5% for CNN and transformer methods. On the REMBRANDT dataset, DA-MLM achieved 92.3% accuracy with segmentation improvements of 3–7% over existing U-Net and expert annotations. Robustness testing indicated 40–60% less degradation under noise, contrast shift, and motion artifacts, and synthetic shifts in scanner location showed negligible performance impairment (<0.06). Cross-domain evaluation also demonstrated 5–11% less degradation than existing methods. Conclusions: In summary, DA-MLM demonstrates improved accuracy, segmentation fidelity, and robustness to perturbations, as well as strong cross-domain generalization indicating the suitability for deployment in multicenter MRI applications where variation in imaging performance is unavoidable.

20 July 2025 in Journal of Advanced Computational Intelligence and Intelligent Informatics
Verified
Journal Article
Open access
MEMohamed Hedi Elhajjej
Mohamed Hedi Elhajjej
SSSalwa Said
Salwa Said
NANouha Arfaoui
Nouha Arfaoui
RERidha Ejbali
Ridha Ejbali
Machine learning, particularly deep learning, is a powerful tool for assisting radiologists in analyzing large volumes of chest X-ray images, significantly accelerating disease diagnosis. However, privacy regulations and data ownership challenges often hinder the centralization of sensitive patient data required for training. Federated learning (FL) addresses these issues by enabling decentralized model training while preserving data confidentiality. This paper introduces a novel FL framework for secure COVID-19 screening using chest X-rays. Our approach incorporates asynchronous communication to overcome delays caused by device heterogeneity and minimizes server-client interactions to reduce network traffic, enhancing scalability. Furthermore, the framework supports heterogeneous client models, ensuring optimized local training. These innovations preserve privacy while achieving performance levels comparable to centralized systems, setting a benchmark for privacy-preserving AI in healthcare.

1 January 2024 in Journal of Electrical and Computer Engineering
Verified
Journal Article
Open access
SSSalwa Said
Salwa Said
SHSalima Hassairi
Salima Hassairi
SYSiwar Yahia
Siwar Yahia
RERidha Ejbali
Ridha Ejbali
MZMourad Zaied
Mourad Zaied
In recent machine learning applications, promising outcomes have emerged through the integration of Deep Learning (DL) and Extreme Learning Machine (ELM) techniques with wavelet networks (WN), leading to high classification accuracy. Researchers have explored various strategies to enhance classifier performance. Among these methods, Deep Wavelet Autoencoders (AE) and the Deep Wavelet ELM algorithm have been extensively applied in diverse image classification domains. This research paper conducts a comparative analysis between these two wavelet AE‐based techniques. The first approach involves the development of a Deep Stacked Sparse Wavelet Autoencoder (DSSWAE), while the second method focuses on creating a Deep Wavelet ELM Autoencoder (DW‐ELM‐AE). To enable a comprehensive comparison, experiments were conducted using different datasets, specifically MNIST and COIL‐20. The results clearly demonstrate that the Deep Stacked Wavelet AE (DSWAE) and DSSWAE algorithms outperform the State‐of‐the‐Art (SOTA) techniques in terms of accuracy efficiency, emphasizing their significant potential for a wide range of image classification applications.

3 June 2023
Verified
Book Chapter
Closed access
PDF file
MEMohamed Hedi Elhajjej
Mohamed Hedi Elhajjej
NANouha Arfaoui
Nouha Arfaoui
SSSalwa Said
Salwa Said
RERidha Ejbali
Ridha Ejbali