HAL Portal IOGS (nstitut d'Optique Graduate School)
Not a member yet
12589 research outputs found
Sort by
FASER: a tool for vectorial point spread function simulation with applications in stimulated emission depletion microscopy
No full textInternational audienceWe introduce FASER, a software package designed to simulate the excitation point spread functions (PSFs) of microscopes. It is written in Python as a plugin for the open-source platform Napari. Using a full-vectorial computational approach to simulate the electromagnetic fields within the focal region makes precise predictions and allows detailed analyses of excitation PSFs. FASER is intended as a pedagogical tool enabling users to explore the impacts of various geometrical and optical parameters of practical importance on the performance of the microscope. It supports the modeling of complex beam profiles, including donut and bottle-shaped beams, which are commonly used in advanced microscopy techniques such as stimulated emission depletion (STED) microscopy. Through specific simulations and accessible illustrations, we showcase FASER’s capabilities in capturing characteristic features of STED microscopy, making it a practical resource for researchers and students in optical microscopy to explore and optimize high-resolution imaging techniques
Impact of scattering variation measured by line-field confocal optical coherence tomography on fluorescence measurement by optical spectroscopy: a study on phantoms and human skin models
No full textInternational audienceIntroduction: Since skin tissues feature highly inter and intra-individual variable scattering properties, it is of interest for fluorescence spectroscopy applied to skin cancer diagnostic assistance to be combined with a device able to measure scattering properties of skin tissues in vivo and further correct fluorescence spectra. This study aims to explore the interest of combining two devices previously used in vivo during clinical trials: line-field confocal optical coherence tomography (LC-OCT) for scattering properties estimation and fluorescence spectroscopy for measuring the modification of endogenous fluorescence induced by carcinogenesis. Methods: This study was performed on liquid phantoms and on commercially available in vitro-grown 3D human skin models. Bulk scattering properties of liquid fluorescent phantoms were estimated separately at 800 nm as a function of Intralipid concentration from LC-OCT images using a model based on the modified Beer-Lambert law. These results were then compared with values obtained with double-integrating spheres and collimated transmittance measurements followed by Inverse Adding-Doubling estimation of bulk scattering properties. Changes in the amplitude of the chlorin-e6 fluorescence peak were measured as a function of IL concentration using fluorescence spectroscopy. The results obtained on phantoms were then validated with the in vitro-grown skin model. Results: Measurements performed on liquid phantoms showed that LC-OCT overestimates scattering coefficient and anisotropy factor by approximately 20-30% compared to values measured by a method (usable only ex vivo) considered here as the gold standard: double integrating spheres-based optical bench. Fluorescence spectroscopy was employed to measure changes in chlorin-e6 fluorescence measured intensity relative to varying Intralipid concentration. Optical characterization of human skin models confirmed their similarity with in vivo human skin in terms of morphology and of autofluorescence signals. LC-OCT was used to detect dermal scattering coefficient increase induced by optical clearing agent, which was opposite to the typically expected decrease with effective optical clearing. It was likely attributable to specific morphological features of the artificial skin that hindered the clearing process, resulting in only hyperosmotic effect. Spectral measurements supported these findings. Conclusion: These findings underscore the interest of combining both optical methods, LC-OCT imaging and autofluorescence spectroscopy, to assess pathology-related fluorophore and scattering alterations in vivo
Operation of Two-axis Tracking Monofacial and Bifacial Photovoltaic Systems in Humid Continental Climate
No full textInternational audienceWith the large-scale deployment of renewable energy, photovoltaic systems are installed worldwide, including in regions with humid continental climate. We present here the energy production of 2-axis tracking systems equipped with either monofacial or bifacial modules. We show that severe weather conditions, especially in winter, force the systems to operate in a degraded mode (i.e. 10° fixed tilt) that leads to yearly losses of ~4%. Other losses that include snow-induced losses lead to 12% and 9% additional losses for monofacial and bifacial systems, respectively. Bifacial modules are better suited than monofacial modules in the studied climate and system configuration, thanks to a higher sensitivity to albedo and higher ability to shed snow during winter, that both amplify the bifaciality gain during snowy months
Conformal Online Learning of Deep Koopman Linear Embeddings
No full textInternational audienceWe introduce Conformal Online Learning of Koopman embeddings (COLoKe), a novel framework for adaptively updating Koopman-invariant representations of nonlinear dynamical systems from streaming data. Our modeling approach combines deep feature learning with multistep prediction consistency in the lifted space, where the dynamics evolve linearly. To prevent overfitting, COLoKe employs a conformal-style mechanism that shifts the focus from evaluating the conformity of new states to assessing the consistency of the current Koopman model. Updates are triggered only when the current model’s prediction error exceeds a dynamically calibrated threshold, allowing selective refinement of the Koopman operator and embedding. Empirical results on benchmark dynamical systems demonstrate the effectiveness of COLoKe in maintaining long-term predictive accuracy while significantly reducing unnecessary updates and avoiding overfitting
Miniaturized optical system for a chip-based cold-atom inertial sensor
No full textInternational audienceWe miniaturized the complex optical system responsible for the cooling, pumping, and imaging of an on-chip-based cold-atom inertial sensor. This optical bench uses bonded miniature optics and includes all the necessary optical functions. The bench has a volume of 35×25×5cm 3 . We developed a laser frequency lock adapted to the optical bench using saturated absorption in a rubidium cell. The entire laser source based on frequency doubling of 1.56 µm fiber lasers, including the control system and the saturated absorption module, fits in a 5 U rack. Using the miniaturized bench, we realized two- and three-dimensional magneto-optical traps for rubidium 87 atoms
System Analysis for a high-precision high-accuracy Astrometric instrument for HWO
No full textInternational audienceThis study presents a comprehensive system analysis for an instrument onboard the Habitable Worlds Observatory (HWO), designed for high-precision, high-accuracy differential astrometry, with the primary scientific goal to determine the mass of Earth-like planets around the nearest Sun-like stars. The analysis integrates the definition of the mission profile, the instrumental concept architecture, and an error budget that breaks down the key contributors to the sub-µas precision required for a single measurement. A portion of this budget addresses photo-center estimation for both the target and calibration stars used in differential astrometry. Other major contributors are related to instrumental control of systematics in the reconstruction of differential angle measurements from pixel data (focal plane calibration) to on sky line of sight (telescope distortion calibration). End-of-mission astrometry requires multiple observations (typically 100) of the same target distributed over the mission lifetime. We assess the mission profile to estimate the fraction of survey time required for astrometric survey to achieve the science objective. The proposed architecture of the instrument concept is derived from error budget and mission constraints based on a large visible detector array composed of an assembly of multiple CMOS sensor chips resulting in an overall gigapixel focal plane. We evaluate the Technology Readiness Level (TRL) and propose a way forward reaching TRL 5 level for key technologies by the Mission Consolidation Review in 2029.</div
Indirectly LED-pumped Nd:glass regenerative amplifier
No full textInternational audienceWe demonstrate the first LED-pumped Nd:glass regenerative amplifier. It delivers 2 ns pulses at 1053 nm with an energy of 3.8 mJ at a repetition rate of 2 Hz. This performance is achieved using a Ce:LuAG luminescent concentrator producing 3.4 kW of optical peak power in the yellow spectral range and pumped by 3200 blue LEDs. This indirect LED pumped technology demonstrates high power-scaling capabilities and shows great potential for high energy Nd:glass laser chains
Realization of a doped quantum antiferromagnet in a Rydberg tweezer array
No full textInternational audienceDoping an antiferromagnetic (AFM) Mott insulator is central to our understanding of a variety of phenomena in strongly correlated electrons, including high-temperature superconductors1,2. To describe the competition between tunnelling t of hole dopants and AFM spin interactions J, theoretical and numerical studies often focus on the paradigmatic t–J model3 and the direct analogue quantum simulation of this model in the relevant regime of high-particle density has long been sought4,5. Here we realize a doped quantum antiferromagnet with next-nearest-neighbour (NNN) tunnellings t′ (refs. 6,7,8,9,10) and hard-core bosonic holes11 using a Rydberg tweezer platform. We use coherent dynamics between three Rydberg levels, encoding spins and holes12, to implement a tunable bosonic t–J–V model allowing us to study previously inaccessible parameter regimes. We observe dynamical phase separation between hole and spin domains for |t/J| ≪ 1 and demonstrate the formation of repulsively bound hole pairs in a variety of spin backgrounds. The interference between NNN tunnellings t′ and perturbative pair tunnelling gives rise to light and heavy pairs depending on the sign of t. Using the single-site control allows us to study the dynamics of a single hole in 2D square lattice (anti)ferromagnets. The model we implement extends the toolbox of Rydberg tweezer experiments beyond spin-1/2 models13 to a larger class of t–J and spin-1 models14,15
Dosimétrie en temps réel : faisabilité et optimisation de capteurs avec fibres optiques radioluminescentes
No full textThis thesis explores the feasibility and optimization of optical fiber-based dosimeters, with a particular focus on their application in radiotherapy, including X-ray and proton beam dosimetry. Optical fibers are promising candidates for real-time, high-spatial-resolution dosimetry due to their compact size, mechanical flexibility, and immunity to electromagnetic interference.Through the ANR FIDELIO project, this work introduces doped sol-gel-derived silica fibers and SPCVD fibers, specifically designed for dosimetric applications, representing a significant advancement in optical fiber dosimetry. The study primarily investigates silica-based optical fibers doped with various elements such as Cerium (Ce), Nitrogen (N), Gadolinium (Gd), and a combination of Ce and Gd, as well as additional dopants like Copper (Cu) and Terbium (Tb) under proton irradiation. The response of these fibers was characterized under both X-ray and proton beams, aiming to evaluate their radioluminescence (RIL) signal stability, sensitivity, and reproducibility. In particular the key aspects are:1.Pre-irradiation EffectsPre-irradiation (up to 1 MGy) significantly enhances the fiber radioluminescence response, leading to increased sensitivity and improved stability. The dose-rate dependence of the fibers becomes more linear after pre-irradiation, facilitating sensor calibration. A significant reduction in 'memory effects' as afterglow and bright burn effects was observed, which improves signal reliability. These improvements are likely due to deep trap filling, which suppresses non-radiative recombination and enhances radiative processes.2. Systematic calibration studies:Emphasizing the importance of systematic calibration for these sensors, the thesis highlighted how a statistical approach provides an estimation of the reliability and reproducibility limits for these dosimeters. While the results are promising for applications requiring a maximum uncertainty of 15-20%, they are still insufficient for medical treatments, where higher accuracy is critical. By the way, it has been observed that a system specifically designed and adapted for optical fiber sensors can significantly reduce these uncertainties by optimizing elements such as connections, splices, the detection and transport system.3.Depth-dose profile measurements under proton irradiationOptical fibers demonstrated the ability to reproduce the Bragg Peak and Spread-Out Bragg Peak (SOBP) curves, making them promising for proton therapy dosimetry. Pre-irradiated fibers, particularly CeTb-doped fibers, exhibited the best match with reference traces.4.Tapered optical fibers for enhanced spatial resolutionReducing the fiber core diameter via tapering increased sensitivity per unit area by a maximum factor of ~130, making them suitable for microdosimetry applications. However, pre-irradiation in tapered fibers led to a significant increase in radiation-induced attenuation (RIA), suggesting the creation of precursor centers that absorb radiation, potentially limiting long-term performance. Monte Carlo Geant4 simulations confirmed that the reduced geometry does not impact the dose absorption under X-ray irradiation, though equilibrium conditions are lost for gamma rays above 500 keV.5.Energy dependence and calibration under proton irradiationThe radioluminescence response of optical fibers showed only a weak (~15%) dependence on proton energy, especially for pre-irradiated fibers. This suggests that a single calibration could be used for a broad energy range (19–73 MeV). Geant4 simulations confirmed that protons in this energy range fully penetrate the fiber, ensuring accurate dosimetric measurements.Cette thèse explore la faisabilité et l’optimisation des dosimètres à base de fibres optiques, en mettant particulièrement l’accent sur leur application en radiothérapie, y compris la dosimétrie des rayons X et des faisceaux de protons. Les fibres optiques sont des candidates prometteuses pour une dosimétrie en temps réel et à haute résolution spatiale, grâce à leur taille compacte, leur flexibilité mécanique et leur immunité aux interférences électromagnétiques. Dans le cadre du projet ANR FIDELIO, ce travail introduit des fibres de silice dopées issues du procédé sol-gel ainsi que des fibres SPCVD, spécialement conçues pour des applications dosimétriques, représentant une avancée significative dans la dosimétrie par fibres optiques. L’étude se concentre principalement sur les fibres de silice dopées avec divers éléments tels que le cérium (Ce), l’azote (N), le gadolinium (Gd) et une combinaison de Ce et Gd, ainsi que des dopants supplémentaires comme le cuivre (Cu) et le terbium (Tb) sous irradiation protonique. La réponse de ces fibres a été caractérisée sous faisceaux de rayons X et de protons, afin d’évaluer la stabilité, la sensibilité et la reproductibilité du signal de radioluminescence (RIL). Les principaux aspects étudiés sont :1. Effets de la pré-irradiation. La pré-irradiation (jusqu’à 1 MGy) améliore considérablement la réponse en radioluminescence des fibres, augmentant leur sensibilité et leur stabilité. La dépendance au débit de dose devient plus linéaire après pré-irradiation, facilitant l’étalonnage des capteurs. Ces améliorations sont probablement dues au remplissage des pièges profonds, supprimant la recombinaison non radiative et renforçant les processus radiatifs.2. Études d’étalonnage systématique. Cette thèse met en avant l’importance d’un étalonnage systématique, soulignant qu’une approche statistique permet d’estimer les limites de fiabilité et de reproductibilité des dosimètres. Les résultats sont prometteurs pour des applications nécessitant une incertitude maximale de 15-20 %, mais restent insuffisants pour les traitements médicaux, où une précision accrue est essentielle. Cependant, un système spécifiquement conçu et adapté aux capteurs à fibres optiques pourrait réduire ces incertitudes en optimisant des éléments tels que les connexions, les soudures, ainsi que le système de détection et de transport du signal. 3. Mesures de profils de dose en profondeur sous irradiation protonique. Les fibres optiques ont démontré leur capacité à reproduire les courbes du pic de Bragg et du Spread-Out Bragg Peak, les rendant prometteuses pour la dosimétrie en protonthérapie. Les fibres pré-irradiées dopées CeTb ont montré la meilleure correspondance avec les traces de référence. 4. Fibres optiques effilées pour une meilleure résolution spatiale. La réduction du diamètre du cœur des fibres par effilement a augmenté la sensibilité par unité de surface d’un facteur maximal de ~130, rendant ces capteurs adaptés aux applications de microdosimétrie. Cependant, la pré-irradiation des fibres effilées a entraîné une augmentation significative de l’atténuation induite par les radiations (RIA), suggérant la formation de centres précurseurs absorbant le rayonnement, ce qui pourrait limiter les performances à long terme. Les simulations Monte Carlo ont confirmé que la géométrie réduite n’impacte pas l’absorption de dose sous irradiation X, bien que les conditions d’équilibre soient perdues pour les rayons gamma au-delà de 500 keV.5. Dépendance énergétique sous irradiation protonique. La réponse en radioluminescence des fibres optiques a montré une faible dépendance (~15 %) à l’énergie des protons, en particulier pour les fibres pré-irradiées. Cela suggère qu’un étalonnage unique pourrait être utilisé sur une large gamme d’énergies (19–73 MeV). Les simulations Geant4 ont confirmé que les protons dans cette gamme d’énergie pénètrent entièrement la fibre, garantissant ainsi des mesures dosimétriques précises
Speech transformer models for extracting information from baby cries
No full textInternational audienceTransfer learning using latent representations from pretrained speech models achieves outstanding performance in tasks where labeled data is scarce. However, their applicability to non-speech data and the specific acoustic properties encoded in these representations remain largely unexplored. In this study, we investigate both aspects. We evaluate five pretrained speech models on eight baby cries datasets, encompassing 115 hours of audio from 960 babies. For each dataset, we assess the latent representations of each model across all available classification tasks. Our results demonstrate that the latent representations of these models can effectively classify human baby cries and encode key information related to vocal source instability and identity of the crying baby. In addition, a comparison of the architectures and training strategies of these models offers valuable insights for the design of future models tailored to similar tasks, such as emotion detection.</div