HAL-UNILIM
Not a member yet
39506 research outputs found
Sort by
La contribution des fondations à l’intérêt général : une perspective sociohistorique
No full textInternational audienc
Hyperspectral image super-resolution based on spectral graph wavelet transform
No full textInternational audienceSuper-resolution techniques are essential for enhancing spatial detail in images, especially hyperspectral data, which possess rich spectral information but commonly suffer from low spatial resolution. Despite advancements in imaging hardware, acquiring high-resolution hyperspectral images remains challenging due to the vast data volume and acquisition limitations. Graph Signal Processing (GSP), and particularly the Spectral Graph Wavelet Transform (SGWT), offers an effective framework for processing signals on irregular domains by capturing the intrinsic relationships between spatial and spectral components through graph models. In this paper, we introduce a novel super-resolution method for hyperspectral images that leverages SGWT to extract wavelet coefficients from the data. An embedding network converts the low-resolution input into discriminative feature maps, which are then used to predict corresponding wavelet coefficient images. These predicted coefficients enable the reconstruction of a high-resolution hyperspectral image via the inverse SGWT. By jointly exploiting spatial and spectral information embedded within the hyperspectral cube, the proposed approach enhances image quality effectively. Experimental results validate the effectiveness of the proposed method
La Cour européenne des droits de l'homme prudente praticienne de la justice climatique
No full textInternational audienc
Anatomical feasibility study on direct coaptation of the lateral femoral cutaneous nerve to the tensor fasciae latae muscle nerve as a treatment for meralgia paresthetica by targeted muscle reinnervation
No full textInternational audienc
Comparison of venous drainage characteristics of non-adhesive liquid embolic agents in brain AVMs: The “Tear-away phenomenon” of PHIL
No full textInternational audienceBackground Hemorrhage is a major complication of brain arteriovenous malformations (AVMs) embolization, which can be related to persistent arteriovenous shunts that were not completely occluded during the embolization. In transvenous embolization (TVE) this risk is deemed higher for AVMs larger than 3 cm featuring multiple veins of drainage. Herein, we will discuss a few selected cases where brain AVMs with more than one draining vein were deemed safe for curative embolization with advanced endovascular techniques after a careful anatomical study through the four dimensional-digital subtraction angiography (4D-DSA) imaging. Methods Four brain AVMs with more than one draining vein were selected from our database. A curative embolization through advanced endovascular techniques with curative intent was planned with the 4D-DSA. Immediate and 6 month angiographic controls are provided, as well as post-procedure MRI. Results All 4 cases were successfully embolized, with stable 6 month angiographic controls. Although multiple draining veins were present, the 4D-DSA allowed the identification of favorable venous anatomy, such as a connection point between draining veins in case 1, as well as a main draining vein in cases 2 and 3. Case four is an exception to this, and although the AVM was cured patient presented asymptomatic ischemia after treatment. The anatomical insights gained through 4D-DSA imaging and the advanced endovascular techniques employed are discussed in detail. Conclusion Brain AVMs with multiple draining veins may behave like featuring a single draining vein, which can be identified through a careful anatomical study in the 4D-DSA, highlighting its importance for a safe endovascular treatment
SmartDetector: a valid and affordable AI-based markerless motion capture system for psychological experiments
No full textInternational audienceIn this paper, we evaluate the validity of SmartDetector—a markerless motion capture device based on artificial intelligence—by comparing it to a conventional optoelectronic motion capture system (i.e., Qualisys) in the context of stimuli construction for psychological experiments. To this end, simultaneous motion capture recordings were performed using both the conventional system and the SmartDetector, enabling the creation of 2D kinematic data and the generation of two versions of point-light displays. Three perceptual experiments were then conducted to facilitate comparisons between the two systems: a recognition task, a detection task, and a discrimination task. Bayesian analysis was employed to test the null hypotheses regarding the stimuli generated by SmartDetector versus those produced by the conventional motion capture system. The results suggest that participants achieved comparable performance across both types of stimuli depending on the task with reduced processing time and lower costs for SmartDetector. These findings appear to validate SmartDetector as a reliable and accessible alternative for creating point-light displays, reducing processing time and costs, thus reinforcing its potential for widespread adoption by professionals in the field
Source‐Connected Field‐Plate Effects on Static Current, Trap Signature, and Thermal Resistance of AlGaN/GaN High Electron Mobility Transistor
No full textInternational audienceThe source‐connected field‐plate (SC‐FP) effects on the electrical, trapping, and thermal characteristics of 150 nm AlGaN/GaN high electron mobility transistors (HEMTs) are investigated, relative to the unterminated HEMT. Static electrical characterization reveals that the SC‐FP structure reduces drain current ( I DS ) and subthreshold leakage current compared to the FP‐free devices, due to the decrease in peak electric field. Alongside, gate leakage current ( I GS ) is considerably mitigated, with the improvement in subthreshold slope. The drain current transient (DCT) and output‐admittance ( Y 22 ) measurements indicate that the emission time constant of the buffer trap prolongs in the SC‐FP HEMT, because of moderated Poole‐Frenkel emission. This results in a marginally higher activation energy for the buffer trap. At the same time, the SC‐FP structure mitigates the electron trapping in the HEMT, as identified from the DCT and Y 22 signal amplitude. A methodology is established to estimate the thermal resistance ( R TH ) directly from the DCT experiments, conducted at different dissipated powers and temperatures. Using this DCT method, the effective R TH of the HEMT with and without SC‐FP is reported. The experimental observations are also substantiated with technology computer‐aided design (TCAD) simulation findings, enabling a comprehensive understanding of the underlying physical mechanisms governing the macroscopic transistor characteristics
« Pour un théâtre du dépassement : l’archipélisation du monde dans Matrices de Daniely Francisque »
No full textInternational audienc
Unlocking the Microbiome-Metabolome Nexus for Innovative One-Health Solution
No full textMicrobial communities, encompassing a vast taxonomic diversity, are fundamental to ecosystem integrity, biogeochemical cycles, and the health of humans, animals, and plants, along the One Health concept. A major scientific goal is to understand how these complex consortia function, interact, and adapt to environmental changes. Microbial meta-metabolomics has emerged as a powerful approach to tackle this by characterizing the collective metabolome of an entire community, linking it to environmental conditions and biogeochemical processes. It captures the functional output of both cultivable and uncultivable organisms, tracing chemical interactions and the impact of environmental perturbations. However, while meta-metabolomics provides a comprehensive snapshot of community chemistry, it alone cannot decipher the precise dynamics of which microorganisms are producing metabolites, when, where, and why. To address this, we propose the Microbial Metabolomics Framework (MiMetWork). This novel framework expands beyond descriptive meta-metabolomics to integrate spatial and temporal metabolomic characterizations with other omics data and phenotyping techniques. MiMetWork employs high-throughput screening of various microbiome components—from single cells to complex communities—under controlled conditions to elucidate ecophysiological functions and interaction mechanisms. By combining untargeted and targeted metabolomic datasets with microbial composition and pathway information, MiMetWork aims to build causal models of microbiome function and adaptation. This review outlines how this integrative framework leverages technological advances to elucidate microbiome interactions and functional responses across human, animal, and environmental niches, thereby addressing critical research gaps and enhancing our predictive understanding of microbiomes within the One Health paradigm