Background/Objectives: Accurate delineation of the optic disc and optic cup in retinal fundus photographs is a prerequisite for automated glaucoma screening. While encoder–decoder segmentation models have advanced considerably, the contribution of upstream preprocessing to segmentation accuracy, and the stability of that contribution across repeated training runs, remain insufficiently characterized. Methods: Five preprocessing pipelines, baseline, Contrast Limited Adaptive Histogram Equalization (CLAHE), Region of Interest (ROI) cropping, ROI+CLAHE, and CLAHE with heavy augmentation, were benchmarked under a fixed EfficientUNet++ model with an EfficientNet-B7 encoder on three publicly available fundus datasets (REFUGE, ORIGA, and Drishti-GS). Every configuration was retrained under three independent random seeds (42, 15, and 89) to assess run-to-run variability. Seed-level standard deviations accompany every reported mean and define the confidence limit on each ranking. Results: On REFUGE, CLAHE with augmentation (Config 5) achieved the strongest mean Dice (disc 0.9523±0.0017; cup 0.8348±0.0018). On ORIGA, all five configurations clustered within 0.0067 disc Dice; ROI+CLAHE (Config 4) was marginally ahead on disc (0.9681±0.0002) and augmentation led on the cup (0.8873±0.0024). On Drishti-GS, all five configurations converged successfully once optimizer and loss settings were corrected; the near-total failures seen in earlier single-run experiments reflected a configuration problem, not the small (81-image) training set. Conclusions: CLAHE applied to full-resolution images is the single most consistently beneficial preprocessing choice across all three datasets. ROI+CLAHE showed a small, initialization-stable advantage on ORIGA, but ROI crop centres were derived from ground-truth centroids, an oracle localization setting, and these results should not be interpreted as achievable by a fully automated pipeline. Data augmentation showed a consistent reduction in initialization sensitivity on small datasets and may be beneficial as a default strategy.
Abdullah Alajmi, Youssef Elnahal, Mohamed Othman et al.· Diagnostics· 0 citations
The proliferation of Internet of Things (IoT) devices in smart home environments has dramatically expanded the attack surface for cyber threats, particularly botnet-driven Distributed Denial of Service (DDoS) attacks. Centralized Intrusion Detection Systems (IDS) are ill-suited to this domain because they violate user privacy, introduce single points of failure, and incur prohibitive communication overhead. Federated Learning (FL) offers a compelling privacy-preserving alternative, yet existing FL-based IDS solutions either deploy convolutional or recurrent models in isolation, lack human-interpretable outputs, or neglect real-world deployability constraints. This paper proposes FedShield-IDS, a novel federated intrusion detection framework that integrates a hybrid one-dimensional Convolutional Neural Network with Long Short-Term Memory units to simultaneously capture spatial traffic fingerprints and long-range temporal attack dynamics across IoT edge devices. Model interpretability is addressed through the integration of SHapley Additive exPlanations (SHAP), enabling administrators to receive human-readable justifications for every detected anomaly. The system is trained and evaluated on the large-scale CICIoT2023 dataset, comprising 712,311 flow records spanning eight attack families including DDoS, DoS, Mirai, Reconnaissance, Spoofing, Injection, and Malware. A multi-stage preprocessing pipeline combining infinite-value imputation, logarithmic feature scaling, Min-Max normalization, temporal windowing, and localized SMOTE oversampling is applied within each federated client to address non-IID data and extreme class imbalance. Federated Averaging aggregates encrypted model updates across seven virtual IoT client nodes over five communication rounds without exchanging raw traffic data, under a formal threat model characterizing the system’s adversarial assumptions and data-confidentiality guarantees. Experimental results demonstrate a Mirai F1-score of 0.99, a DDoS precision of 0.97, and a global weighted F1-score of 0.76 across all eight classes. Comprehensive kernel-size, architecture, and preprocessing ablations confirm the necessity of each design choice, and independent cross-dataset evaluation on the Edge-IIoTset benchmark achieves 98.58% accuracy, demonstrating strong generalization beyond CICIoT2023. The framework achieves sub-500 ms threat mitigation, empirically confirmed via a mitigation-gate threshold sensitivity analysis, and generates SHAP-gated explanations for every alert, bridging the gap between high-accuracy detection and the transparency required for trustworthy smart-home security.
Ghada Abdelhady, Karim Wael Hussein, Islam Anwar Ali Gad· Scientific Reports· 0 citations
It is suggested that effective teacher preparation must transcend content delivery and focus on equipping teachers with adaptive, student-centred methodologies grounded in cognitive science.
A. Al-Zahrani, Ghada Abdelhady· International Research Journ...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
