Recent developments underscore the potential of textual information in enhancing learning models for a deeper understanding of medical visual semantics. However, language-guided medical image segmentation still faces a challenging issue. Previous works employ implicit architectures to embed textual information. This leads to segmentation results that are inconsistent with the semantics represented by the language, sometimes even diverging significantly. To this end, we propose a novel cross-modal conditioned Reconstruction for Language-guided Medical Image Segmentation (RecLMIS) to explicitly capture cross-modal interactions, which assumes that well-aligned medical visual features and medical notes can effectively reconstruct each other. We introduce conditioned interaction to adaptively predict patches and words of interest. Subsequently, they are utilized as conditioning factors for mutual reconstruction to align with regions described in the medical notes. Extensive experiments demonstrate the superiority of our RecLMIS, surpassing LViT by 3.74% mIoU on the MosMedData+ dataset and 1.89% mIoU on the QATA-CoV19 dataset. More importantly, we achieve a relative reduction of 20.2% in parameter count and a 55.5% decrease in computational load. The code will be available at https://github.com/ShawnHuang497/RecLMIS.
The purpose of this paper was to analyze molecular weight (MW) distribution in different drinking water treatment processes, and to find out the relationship between dissolved organic carbon (DOC) and disinfection by-products formation potential (DBPFP) in treated water. The results showed that in conventional water treatment, compared with similar methods, namely chlorination (Cl2) and potassium permanganate (KMnO4), pre-ozonation can reduce DOC concentration for larger MW fractions (>30k, 10k–30k and 3k–10 kDalton (Da)), while the smaller MW fraction portion increased. Coagulation, sedimentation and sand-filtration were effective to remove DOC in the larger MW fractions. Quantitative data were statistically explained. In combination with post-ozonation, biological activated carbon (BAC) can eliminate a large amount of DOC in the <1 kDa MW fraction. BAC with a 3-month service life had the optimal absorption and biodegradation effects. The treatment process was better at removing trihalomethane formation ...
26 February 2026 in Journal of Hazardous Materials
Deep-sea aquaculture platforms integrated with offshore renewable energy systems operate under highly dynamic and harsh marine conditions, where wave-induced operational constraints and energy disruptions caused by extreme weather pose major challenges. To address prolonged energy interruptions and constrained operational windows during typhoon events, a rolling optimization framework incorporating wave-state constraints and resource prepositioning is proposed. A sea-state-dependent operational window model is formulated to quantify the impact of wave height on primary loads including feeding, lifting, and net-cleaning tasks, embedding these limits as hard scheduling constraints. An environmental correction model is further developed to quantify the output derating of wind turbines, photovoltaic arrays, and diesel generators under high-humidity and salt-spray exposure. Based on these models, a rolling scheduling strategy adapting to typhoon evolution is designed. During the pre-disaster window, proactive resource prepositioning shifts deferrable loads and increases storage state-of-charge, enabling a transition from economic optimization to resilience-oriented operation. Simulation results demonstrate that the proposed method avoids task violations induced by harsh sea states. Under extreme typhoon conditions, compared with conventional scheduling without sea-state constraints and resource prepositioning, our method cuts total operating cost by 10.1% and essential monitoring load shedding by 78.3%. This strategy improves the safe and stable operation of deep-sea aquaculture platforms, and supports the industry’s sustainable development.