Abstract
Accurate retrieval of post-surgical images plays a critical role in surgical planning for breast cancer patients. However, current content-based image retrieval methods face challenges related to limited interpretability, poor robustness to image noise, and reduced generalization across clinical settings. To address these limitations, we propose a multistage retrieval pipeline integrating saliency-based explainability, noise-reducing image pre-processing, and ensemble learning. Evaluated on a dataset of post-operative breast cancer patient images, our approach achieves contrastive accuracy of 77.67% for Excellent/Good and 84.98% for Fair/Poor outcomes, surpassing prior studies by 8.37% and 11.80%, respectively. Explainability analysis provided essential insight by showing that feature extractors often attend to irrelevant regions, thereby motivating targeted input refinement. Ablations show that expanded bounding box inputs improve performance over original images, with gains of 0.78% and 0.65% contrastive accuracy for Excellent/Good and Fair/Poor, respectively. In contrast, the use of segmented images leads to a performance drop (1.33% and 1.65%) due to the loss of contextual cues. Furthermore, ensemble learning yielded additional gains of 0.89% and 3.60% over the best-performing single-model baselines. These findings underscore the importance of targeted input refinement and ensemble integration for robust and generalizable image retrieval systems. © 2025 Elsevier B.V., All rights reserved.

Abstract
This article introduces Friday, a Mobile Manipulator (MoMa) solution designed at iiLab - INESC TEC. Friday is versatile and applicable in various contexts, including warehouses, naval shipyards, aerospace industries, and production lines. The robot features an omnidirectional platform, multiple grippers, and sensors for localisation, safety, and object detection. Its modular hardware and software system enhances functionality across different industrial scenarios. The system provides a stable platform supporting scientific advancements and meeting modern industry demands, with results verified in the aerospace, automotive, naval, and logistics.

Abstract
Breast segmentation has a critical role for objective pre and postoperative aesthetic evaluation but challenged by limited data (privacy concerns), class imbalance, and anatomical variability. As a response to the noted obstacles, we introduce an encoder–decoder framework with a Segment Anything Model (SAM) backbone, enhanced with synthetic depth maps and a multiterm loss combining weighted crossentropy, convexity, and depth alignment constraints. Evaluated on a 120patient dataset split into 70% training, 10% validation, and 20% testing, our approach achieves a balanced test dice score of 98.75%—a 4.5% improvement over prior methods—with dice of 95.5% (breast) and 89.2% (nipple). Ablations show depth injection reduces noise and focuses on anatomical regions, yielding dice gains of 0.47% (body) and 1.04% (breast). Geometric alignment increases convexity by almost 3% up to 99.86%, enhancing geometric plausibility of the nipple masks. Lastly, crossdataset evaluation on CINDERELLA samples demonstrates robust generalization, with small performance gain primarily attributable to differences in annotation styles. © 2025 Elsevier B.V., All rights reserved.

Abstract
Stroke often leads to severe motor impairment, especially in the upper limbs, greatly reducing a patient's ability to perform daily tasks. Effective rehabilitation is essential to restore function and improve quality of life. Traditional therapies, while useful, may lack engagement, leading to low motivation and poor adherence. Gamification-using game-like elements in non-game contexts-offers a promising way to make rehabilitation more engaging. The authors explore a gamified rehabilitation system designed in Unity 3D using a Kinect V2 camera. The game includes key features such as adjustable difficulty, real-time and predominantly positive feedback, user friendliness, and data tracking for progress. The evaluations were conducted with 18 healthy participants, most of whom had prior virtual reality experience. About 77% found the application highly motivating. While the gameplay was well received, the visual design was noted as lacking engagement. Importantly, all users agreed that the game offers a broad range of difficulty levels, making it accessible to various users. The results suggest that the system has strong potential to improve rehabilitation outcomes and encourage long-term use through enhanced motivation and interactivity.

Abstract

Abstract
The increasing reliance on Networked Microgrids (NMGs) for decentralized energy management introduces unprecedented cybersecurity risks, particularly in the context of False Data Injection Attacks (FDIA). While traditional FDIA studies have primarily focused on network-based intrusions, this work explores a novel cyber-physical attack vector leveraging Unmanned Aerial Vehicles (UAVs) to execute sophisticated cyberattacks on microgrid operations. UAVs, equipped with communication jamming and data spoofing capabilities, can dynamically infiltrate microgrid communication networks, manipulate sensor data, and compromise power system stability. This paper presents a multi-objective optimization framework for UAV-assisted FDIA, incorporating Non-dominated Sorting Genetic Algorithm III (NSGA-III) to maximize attack duration, disruption impact, stealth, and energy efficiency. A comprehensive mathematical model is formulated to capture the intricate interplay between UAV operational constraints, cyberattack execution, and microgrid vulnerabilities. The model integrates flight path optimization, energy consumption constraints, signal interference effects, and adaptive attack strategies, ensuring that UAVs can sustain long-duration cyberattacks while minimizing detection risk. Results indicate that UAV-assisted cyberattacks can induce power imbalances of up to 15%, increase operational costs by 30%, and cause voltage deviations exceeding 0.10 p.u.. Furthermore, analysis of attack success rates vs. detection mechanisms highlights the limitations of conventional rule-based anomaly detection, reinforcing the need for adaptive AI-driven cybersecurity defenses. The findings underscore the urgent necessity for advanced intrusion detection systems, UAV tracking technologies, and resilient microgrid architectures to mitigate the risks posed by airborne cyber threats.