INRIA a CCSD electronic archive server
Not a member yet
122212 research outputs found
Sort by
Leveraging Deep Learning for Immune Cell Quantification and Prognostic Evaluation in Radiotherapy-Treated Oropharyngeal Squamous Cell Carcinomas.
No full textInternational audienceThe tumor microenvironment (TME) plays a critical role in cancer progression and therapeutic responsiveness, with the tumor immune microenvironment (TIME) being a key modulator. In head and neck squamous cell carcinomas (HNSCC), immune cell infiltration significantly influences the response to radiotherapy (RT). A better understanding of the TIME in HNSCC could help identify patients most likely to benefit from combining RT with immunotherapy. Standardized, cost-effective methods for studying TIME in HNSCC are currently lacking. This study aims to leverage deep learning (DL) to quantify immune cell densities using immunohistochemistry (IHC) in untreated oropharyngeal squamous cell carcinoma (OPSCC) biopsies of patients scheduled for curative RT, and to assess their prognostic value. We analyzed 84 pre-treatment formalin-fixed paraffin-embedded (FFPE) tumor biopsies from OPSCC patients. Immunohistochemistry was performed for CD3, CD8, CD20, CD163, and FOXP3, and whole slide images (WSIs) were digitized for analysis using a U-Net-based DL model. Two quantification approaches were applied: a cell-counting method and an area-based method. These methods were applied to stained regions. The DL model achieved high accuracy in detecting stained cells across all biomarkers. Strong correlations were found between our DL pipeline, the HALO® Image Analysis Platform, and the open-source QuPath software for estimating immune cell densities. Our DL pipeline provided an accurate and reproducible approach for quantifying immune cells in OPSCC. The area-based method demonstrated superior prognostic value for recurrence-free survival (RFS), when compared to the cell-counting method. Elevated densities of CD3, CD8, CD20, and FOXP3 were associated with improved RFS, while CD163 showed no significant prognostic association. These results highlight the potential of DL in digital pathology for assessing TIME and predicting patient outcomes. Further validation in larger cohorts is recommended
Approaching Immediate Total Thyroidectomy in First-Side Loss of Neuromonitoring Signal With Caution
No full textInternational audienc
ASO Author Reflections: Complete Pathological Response after Neoadjuvant Treatment for Pancreatic Ductal Adenocarcinomas—Curative Surgery in Cured Patients?
No full textInternational audienc
Convergence rates for an Adaptive Biasing Potential scheme from a Wasserstein optimization perspective
No full text52 pagesFree-energy-based adaptive biasing methods, such as Metadynamics, the Adaptive Biasing Force (ABF) and their variants, are enhanced sampling algorithms widely used in molecular simulations. Although their efficiency has been empirically acknowledged for decades, providing theoretical insights via a quantitative convergence analysis is a difficult problem, in particular for the kinetic Langevin diffusion, which is non-reversible and hypocoercive. We obtain the first exponential convergence result for such a process, in an idealized setting where the dynamics can be associated with a mean-field non-linear flow on the space of probability measures. A key of the analysis is the interpretation of the (idealized) algorithm as the gradient descent of a suitable functional over the space of probability distributions
Améliorer la reconstruction du génome microbien dans des environnements complexes en combinant le séquençage à lecture courte et à lecture longue
No full textSoil is one of the most diverse microbial ecosystems, yet a significant portion of its microbial "dark matter" remains uncharacterised. Sequencing technologies have improved our understanding, but challenges persist in comprehensively characterising soil microbial diversity. In this study, we employed PacBio HiFi long-read (LR) and Illumina short-read (SR) whole-genome sequencing (WGS) to reconstruct metagenome-assembled genomes (MAGs) from a soil sample. Metabarcoding analyses complemented our approach by assessing microbial diversity and evaluating the proportion of taxa captured by WGS. Our results demonstrate that LR sequencing significantly enhances genome contiguity and completeness compared to SR methods, which yield more fragmented assemblies. By integrating SR and LR data, we improved binning accuracy, leading to a more precise taxonomic resolution of soil microbial diversity. While long-read sequencing provides the most comprehensive WGS representation, our findings highlight that low-abundant microbial taxa remain undetected due to sequencing depth limitations.Le sol est l'un des écosystèmes microbiens les plus diversifiés, mais une grande partie de ces communautés reste inexplorée. Les technologies de séquençage ont amélioré notre compréhension, mais des défis subsistent pour caractériser pleinement cette diversité microbienne. Dans cette étude, nous avons utilisé le séquençage à longues lectures PacBio HiFi (LR) et de à lectures courtes Illumina (SR) en séquençage de génome entier (WGS) afin de reconstruire des génomes métagénomiques assemblés (MAGs) à partir d’un échantillon de sol. Des analyses de métabarcoding ont complété notre approche en évaluant la diversité microbienne et en quantifiant la proportion de taxons capturés par le WGS. Nos résultats montrent que le séquençage LR améliore considérablement la continuité et la complétude des génomes par rapport aux méthodes SR, qui produisent des assemblages plus fragmentés. L’intégration des données SR et LR a permis d’accroître la précision du binning, conduisant à une résolution taxonomique plus fine de la diversité microbienne du sol. Bien que le séquençage à lectures longues fournisse une meilleure représentation de la diversité du sol, nos résultats soulignent que certains taxons microbiens peu abondants restent non détectés en raison des limitations de la profondeur de séquençage
Sensorimotor Learning With Stability Guarantees via Autonomous Neural Dynamic Policies
No full textInternational audienceState-of-the-art sensorimotor learning algorithms, either in the context of reinforcement learning or imitation learning, offer policies that can often produce unstable behaviors, damaging the robot and/or the environment. Moreover, it is very difficult to interpret the optimized controller and analyze its behavior and/or performance. Traditional robot learning, on the contrary, relies on dynamical system-based policies that can be analyzed for stability/safety. Such policies, however, are neither flexible nor generic and usually work only with proprioceptive sensor states. In this work, we bridge the gap between generic neural network policies and dynamical system-based policies, and we introduce Autonomous Neural Dynamic Policies (ANDPs) that: (a) are based on autonomous dynamical systems, (b) always produce asymptotically stable behaviors, and (c) are more flexible than traditional stable dynamical system-based policies. ANDPs are fully differentiable, flexible generic-policies that accept any observation input, while ensuring asymptotic stability. Through several experiments, we explore the flexibility and capacity of ANDPs in several imitation learning tasks including experiments with image observations. The results show that ANDPs combine the benefits of both neural network-based and dynamical systembased methods
Averaged Steklov Eigenvalues, Inside Outside Duality and Application to Inverse Scattering
No full textInternational audienceWe introduce a new family of artificial backgrounds corresponding to averaged impedance boundary conditions formulated in an abstract framework. These backgrounds are used to define a finite number of averaged Steklov eigenvalues, which are associated with inverse scattering problems from inhomogeneous media. We prove that these special eigenvalues can be determined from full-aperture, fixed-frequency far-fields using the inside-outside duality method. We then show and numerically demonstrate how this method can be used to reconstruct averaged values of the refractive index.</div
CUTECat: Concolic Execution for Computational Law
No full textInternational audienceMany legal computations, including the amount of tax owed by a citizen, whether they are eligible to social benefits, or the wages due to civil state servants, are specified by computational laws. Their application, however, is performed by expert computer programs intended to faithfully transcribe the law into computer code. Bugs in these programs can lead to dramatic societal impact, e.g., paying employees incorrect amounts, or not awarding benefits to families in need.To address this issue, we consider concolic unit testing, a combination of concrete execution with SMT-based symbolic execution, and propose CUTECat, a concolic execution tool targeting implementations of computational laws. Such laws typically follow a pattern where a base case is later refined by many exceptions in following law articles, a pattern that can be formally modeled using default logic. We show how to handle default logic inside a concolic execution tool, and implement our approach in the context of Catala, a recent domain-specific language tailored to implement computational laws. We evaluate CUTECat on several programs, including the Catala implementation of the French housing benefits and Section 132 of the US tax code. We show that CUTECat can successfully generate hundreds of thousands of testcases covering all branches of these bodies of law. Through several heuristics, we improve CUTECat's scalability and usability, making the testcases understandable by lawyers and programmers alike. We believe CUTECat paves the way for the use of formal methods during legislative processes
Security of differential phase shift QKD from relativistic principles
No full textPublished version with slightly expanded introductionInternational audienceThe design of quantum protocols for secure key generation poses many challenges: On the one hand, they need to be practical concerning experimental realisations. On the other hand, their theoretical description must be simple enough to allow for a security proof against all possible attacks. Often, these two requirements are in conflict with each other, and the differential phase shift (DPS) QKD protocol exemplifies these difficulties: It is designed to be implementable with current optical telecommunication technology, which, for this protocol, comes at the cost that many standard security proof techniques do not apply to it. After about 20 years since its invention, this work presents the first full security proof of DPS QKD against general attacks, including finite-size effects. The proof combines techniques from quantum information theory, quantum optics, and relativity. We first give a security proof of a QKD protocol whose security stems from relativistic constraints. We then show that security of DPS QKD can be reduced to security of the relativistic protocol. In addition, we show that coherent attacks on the DPS protocol are, in fact, stronger than collective attacks. Our results have broad implications for the development of secure and reliable quantum communication technologies, as they shed light on the range of applicability of state-of-the-art security proof techniques
Removing self-intersections in 3D meshes while preserving floating-point coordinates
No full textWe present a straightforward and robust method for resolving the mesh intersection problem. We focus specifically on the challenge caused by the intersections resulting from the conversion of the vertices coordinates from their exact mathematical values to a fixed-precision floatingpoint format. Our method outputs intersection-free models whose vertices coordinates are all represented with double-precision floating-point format. We evaluated our approach thoroughly, considering a large collection of meshes. In particular, we can process all the 4 524 models in Thingi10K [Zhou and Jacobson, 2016] that contain self-intersections. This outperforms previous state-of-the-art approaches: On the 527 models of Thingi10K for which naive rounding fails, Zhou et al. [2016]'s approach is capable of handling 91% of them, and Valque [2024]'s 94%. In terms of time efficiency, our approach handles about 17k vertices per second (serial) on average, which is faster to that of Zhou et al. [2016] by a factor 1.7 on these non-trivial models and is faster than that of Valque [2024] by several order of magnitude.</div