HAL Portal UTC Université de Technologie de Compiègne
Not a member yet
11652 research outputs found
Sort by
Bacterial alginate-based hydrogels reinforced with polysaccharide nanocrystals for water and wastewater decontamination
No full textInternational audienc
Deciphering the role of ergosterol and sphingolipids in the antifungal mode of action of rhamnolipids on Sclerotiniaceae.
No full textInternational audienceRhamnolipids (RLs) are natural glycolipids very promising to reduce the use of pesticides to protect crops. They have efficient antifungal properties against Sclerotiniaceae fungi. They are membranotropic compounds permeabilizing fungal cells. We have previously proposed that a lower ergosterol amount could be linked to a lower antifungal sensitivity. The main objective of this paper was to gain insights into the role of plasma membrane lipids in the antifungal mode of action of RLs on Sclerotiniaceae. We obtained ergosterol biosynthesis pathway mutants of Botrytis cinerea and evaluated their sensitivity to RLs. We also analyzed the sphingolipid contents of B. cinerea and S. sclerotiorum and assessed their involvement by a pharmacological approach. We confirm here that a decrease of the ergosterol content in B. cinerea mutants triggers a lower antimycelial activity of RLs. Cotreatments of both mycelium with the polyene natamycin, which specifically binds to ergosterol, show that the ergosterol role in RL activity involves an access of RLs to this lipid. Conversely, we demonstrate that ceramides and acidic glycosphingolipids could have a protective role against the permeabilization property of RLs. These results demonstrate in vivo the antagonistic role of two major membrane lipids in the mode of action of RLs. They also show that the effective use of RLs as antifungal agents within biocontrol strategies requires consideration of differences in the lipid composition of fungi
Bioactive Compounds from Dandelion (Taraxacum officinale): Advances in Extraction Techniques and Applications
No full textInternational audienceTaraxacum is a genus of flowering plants comprising species commonly known as dandelions. All parts of the dandelion (flowers, stems, roots, and leaves) contain valuable bioactive compounds, including flavonoids, amino, fatty, organic, and phenolic acids, coumarins, lignans, polysaccharides, phytosterols, terpenes, glycoproteins, oligosaccharides, and alkaloids. Dandelion extracts represent a promising feedstock for diverse applications across the food, biomedical, and pharmaceutical industries. The extraction of bioactive compounds from dandelion is essential to access its therapeutic properties, with different techniques used to isolate its various phytochemicals. This review provides a comprehensive overview of recent advances in the application of various techniques for the extraction of bioactive compounds from dandelion. Both conventional and innovative extraction techniques are discussed, with particular emphasis on their respective advantages and limitations
GEOSUR -Lot3 -Test Case Generation for Geolocation Fusion Algorithms
No full textA key component of autonomous agricultural robots is the geolocation system, which relies on multiple sensors (e.g., GNSS -Global Navigation Satellite System -and IMU -Inertial Measurement Unit) combined using fusion algorithms. These algorithms must handle diverse situations caused by environmental conditions, sensor failures, or measurement inaccuracies. One major challenge is to systematically explore these situations with diverse scenarios, in order to generate test cases for validating such algorithms. This paper addresses this challenge by proposing a comprehensive framework that integrates a geolocation fault model with an open-source test case generator (TAF). An experiment conducted on a standard fusion algorithm demonstrates that the proposed framework can generate valid and diverse test cases with very low resource consumption
Adaptive federated control: An event-driven MARL framework for fair and efficient traffic management
No full textInternational audienceNext-generation networks require distributed traffic management to handle dynamic loads, but frequent interagent coordination consumes scarce bandwidth. We propose an adaptive federated multi-agent reinforcement learning (Fed-MARL) framework that triggers model synchronization only when congestion is near. Our congestion index combines queue occupancy, latency, and utilization to detect network stress. When the threshold is exceeded, federated learning aggregates distributed agent models using FedAdam. For fairness, we introduce empathy-weighted reward shaping, where agents balance individual rewards with peer performance, aligning selfish routing with system-wide welfare. DDPG agents deployed at edge switches make routing and load balancing decisions. Evaluated on Fat-Tree K=4 topology, Fed-MARL achieves 93% latency reduction vs. RL-MR (7.77 vs 105 ms), 57% vs. DRAMA (18.15 ms), 192.6 Mbps throughput, and perfect delivery with minimal communication overhead
Novel multi-level optimization-driven 2D/3D matching for reconstructing 3D fetal postures and motion from childbirth MRI during vaginal delivery
No full textInternational audienceBACKGROUND AND OBJECTIVE: The description of 3D fetal postures and movements during vaginal delivery is fundamental for a better understanding of the physiological delivery and the prediction of potential complications and related preventive actions. Recently, 2D dynamic MRI has been developed to characterize in vivo childbirth. However, only 2D information is available. The objective of the present study was to reconstruct the 3D fetal postures and movements from 2D dynamic MRI during vaginal delivery. METHODS: A multi-level optimization-driven 2D/3D matching procedure was developed and evaluated. Manual segmentation from 2D dynamic MRI was performed. A 3D generic fetal template model was used to match each segmented MRI slice. Fetal postures and associated motion trajectories were extracted and evaluated. RESULTS: Consistent results were obtained between the MRI images, segmented slices, and reconstructed 3D fetal postures. A maximum 36.20-degree head-neck extension angle was estimated. A range of torso-neck angles from 1.04 to -16.07 degrees was estimated during the vaginal delivery. CONCLUSIONS: We proposed a proof-of-concept study of a multi-level optimization-driven 2D/3D matching scheme to reconstruct 3D fetal postures and associated kinematic patterns from dynamic MRI data during vaginal delivery simulation. It is expected that this novel dataset will make a potential contribution to the future model development and evaluation of the childbirth process
An analytic decomposition approach to modeling evanescent transient plane waves with an illustrative application to acoustics at a fluid–fluid interface
No full textInternational audienceThis work introduces an analytic decomposition approach for transient plane waves, broadly applicable to their interaction with layered media in linear physics. The time-domain signal, assumed to be analytic, is extended into the complex plane and then analytically decomposed into two parts, in a manner reminiscent of the Wiener–Hopf technique. The method directly provides the Hilbert transform of the signal and the expressions of the complex fields without resorting to Fourier transforms or the calculation of singular integrals. It applies to a wide class of functions capable of simulating most realistic signals, such as multi-frequency oscillatory signals of limited duration, and is particularly relevant whenever certain fields become evanescent. An illustrative application to acoustics at a fluid–fluid interface offers deeper physical insight than previously available, particularly regarding the processes that generate precursor and successor phenomena in the reflected and transmitted fields, taking advantage of both the simplicity and the effectiveness of the method
Convergence of a scheme for a two dimensional nonlocal system of transport equations
No full textIn this paper, we numerically study a two-dimensional system modeling the dynamics of dislocation densities. This system is hyperbolic, but not strictly hyperbolic, and couples two non-local transport equations. It is characterized by weak regularity in both the velocity and the initial data. We propose a semi-explicit finite difference (IMEX) numerical scheme for the discretization of this system, after regularizing the singular velocity using a Fejér kernel. We show that this scheme preserves, at the discrete level, an entropy estimate on the gradient, which then allows us to establish the convergence of the discrete solution to the continuous solution. To our knowledge, this is the first convergence result obtained for this type of system. We conclude with some numerical illustrations highlighting the performance of the proposed scheme
Administering a bioprotective strain to live fish in aquaculture prevents growth of Listeria and spoilage bacteria on the processed fillets
No full textInternational audienceSeafood is prone to rapid spoilage and transmission of foodborne diseases. Here we investigate the effect of the bioprotective strain Carnobacterium divergens V41 on the health of on-growing Atlantic salmon (Salmo salar) and spoilage of its fillets throughout storage. The strain was administered throughout a three-month growth period and applied to the water in the fish tanks prior to harvest to assess its impact on animal health, shelf-life properties of the processed fillets, and growth inhibition of Listeria. Administering the strain through feed for three months had no adverse effect on the farmed fish, and no significant differences were detected in growth or mortality between the treatment and control groups. However, adding C. divergens V41 to the tank water prior to harvesting markedly changed the fillet microbiome, suppressing the growth of spoilage organisms and thereby possibly extending the shelf-life period. A Listeria challenge test demonstrated that exposure of the fish to C. divergens V41 prior to harvesting and processing resulted in over 99.9 % inhibition of Listeria growth on the fillets over a 21-day period, compared with the control group. This is the first study to show that administering a bioprotective strain to live farmed fish can have a bioprotective effect on fish fillets after processing. This is a simple method to modify the food microbiome of a rapidly perishable product without specialised equipment, thereby increasing food safety and reducing food waste
Intégration éthique de l'IA en radiologie en France : perceptions, pratiques et recommandations
No full textArtificial intelligence is increasingly used in radiology, particularly for diagnostic support. While it provides significant clinical benefits, its integration nonetheless raises major ethical and regulatory challenges. These issues notably concern the transparency of algorithms and of the decisions produced by AI systems, human oversight of generated results in order to ensure effective medical control, liability in the event of diagnostic errors involving an AI system, the training of healthcare professionals in the use, limitations, and biases of AI tools, as well as informing patients about the use of AI in their care, the purposes of data processing, and their associated rights. This work is based on a field study combining questionnaires administered to radiologists and patients in order to assess their perceptions, levels of trust, and understanding of the use of AI in radiology, as well as interviews conducted with application engineers to understand the technical, organizational, and regulatory constraints of AI systems and current user support practices. This approach is complemented by an analysis of the main regulatory and ethical frameworks governing AI in healthcare (the AI act, CNIL recommendations, and WHO guidelines). The results highlight a gap between formalized ethical and regulatory requirements (transparency, human oversight, patient information, accountability, and bias management) and real-world practices observed in the field, particularly with regard to patient information, traceability of AI use, and professional training. Based on this crossanalysis of data from questionnaires, interviews, and regulatory texts, concrete operational proposals were formulated, such as the implementation of human oversight protocols, strengthening user training, and improving patient information, in order to promote an ethical and responsible integration of AI in radiology.L’intelligence artificielle est de plus en plus utilisée en radiologie, notamment pour l’aide au diagnostic. Si elle apporte des bénéfices cliniques significatifs, son intégration soulève toutefois des enjeux éthiques et réglementaires majeurs. Ceux-ci concernent en particulier la transparence des algorithmes et des décisions produites par les systèmes d’IA, la supervision humaine des résultats générés afin de garantir un contrôle médical effectif, la responsabilité en cas d’erreur diagnostique impliquant un système d’IA, la formation des professionnels de santé à l’utilisation, aux limites et aux biais des outils d’IA, ainsi que l’information des patients sur l’usage de l’IA dans leur prise en charge, les finalités du traitement de leurs données et leurs droits associés. Ce travail repose sur une enquête terrain combinant des questionnaires soumis à des radiologues et des patients afin d’évaluer leur perception, leur niveau de confiance et leur compréhension de l’usage de l’IA en radiologie, ainsi que des entretiens menés avec des ingénieurs d’application dans le but de comprendre les contraintes techniques, organisationnelles et réglementaires des systèmes d’IA, ainsi que les pratiques actuelles d’accompagnement des utilisateurs. Cette approche est complétée par l’analyse des principaux cadres réglementaires et éthiques encadrant l’IA en santé (AI Act, recommandations de la CNIL et lignes directrices de l’OMS). Les résultats mettent en évidence un écart entre les exigences éthiques et réglementaires formalisées (transparence, supervision humaine, information du patient, responsabilité et gestion des biais) et les pratiques réelles observées sur le terrain, notamment en matière d’information des patients, de traçabilité de l’usage de l’IA et de formation des professionnels. À partir de cette analyse croisée des données issues des questionnaires, des entretiens et des textes réglementaires, des propositions opérationnelles concrètes ont été formulées, telles que la mise en place de protocoles de supervision humaine, le renforcement de la formation des utilisateurs et l’amélioration de l’information des patients, afin de favoriser une intégration éthique et responsable de l’IA en radiologie