A5053383278- Galaxies: Formation, Evolution, Phenomena
- Cosmology and Gravitation Theories
- Medical Imaging Techniques and Applications
- Astronomy and Astrophysical Research
- Radio Astronomy Observations and Technology
- Advanced MRI Techniques and Applications
- Dark Matter and Cosmic Phenomena
- Stellar, planetary, and galactic studies
- Astrophysics and Star Formation Studies
- Scientific Research and Discoveries
- Radiomics and Machine Learning in Medical Imaging
- Advanced X-ray and CT Imaging
- Advanced Vision and Imaging
- Astrophysics and Cosmic Phenomena
- Gamma-ray bursts and supernovae
- Adaptive optics and wavefront sensing
- Sparse and Compressive Sensing Techniques
- Medical Image Segmentation Techniques
- Astronomical Observations and Instrumentation
- Advanced Image Processing Techniques
- Remote Sensing in Agriculture
- Impact of Light on Environment and Health
- Particle physics theoretical and experimental studies
- Image and Signal Denoising Methods
- Geophysics and Gravity Measurements
CEA Paris-Saclay
2016-2025
Commissariat à l'Énergie Atomique et aux Énergies Alternatives
2016-2025
Institut d'Imagerie Biomédicale
2006-2025
Astrophysique, Instrumentation et Modélisation
2014-2024
Centre National de la Recherche Scientifique
2015-2024
Université Paris-Saclay
2017-2024
Inserm
2021-2024
Université Paris Cité
2013-2022
Délégation Paris 7
2014-2021
Sorbonne Paris Cité
2017-2021
Brain PET in small structures is challenged by low resolution inducing bias the activity measurements. Improved spatial may be obtained using dedicated tomographs and more comprehensive modeling of acquisition system during reconstruction. In this study, we assess impact (RM) reconstruction on image quality estimates biologic parameters a clinical study performed high-resolution research tomograph. Methods: An accelerated list-mode ordinary Poisson ordered-subset expectation maximization...
We present first results from the third GRavitational lEnsing Accuracy Testing (GREAT3) challenge, in a sequence of challenges for testing methods inferring weak gravitational lensing shear distortions simulated galaxy images. GREAT3 was divided into experiments to test three specific questions, and included space- ground-based data with constant or cosmologically varying fields. The simplest (control) experiment parametric galaxies realistic distribution signal-to-noise, size, ellipticity,...
A fully 4D joint-estimation approach to reconstruction of temporal sequences 3D positron emission tomography (PET) images is proposed. The method estimates both a set basis functions and the corresponding coefficient for each function at spatial location within image. joint estimation performed through version maximum likelihood expectation maximization (ML-EM) algorithm in conjunction with two different models mean Poisson measured data. first model regards coefficients as unknown...
We present a novel estimate of the cosmological microwave background (CMB) map by combining two latest full-sky surveys: WMAP nine-year and Planck PR1. The joint processing benefits from recently introduced component separation method coined "local-generalized morphological analysis'' (LGMCA) based on sparse distribution foregrounds in wavelet domain. proposed estimation procedure takes advantage IRIS 100 micron as an extra observation galactic center for enhanced dust removal. show that...
The implementation and the measurement of an approximate image based model ECAT HRRT PET scanner response function, designed for its regular OSEM reconstruction software, are presented. system matrix used in iterative is factorized into two terms: first a modeling blurring effects space, followed by projection matrix. methodology to measure elements presented applied three scanners. A spatially invariant resolution was chosen; columns then defined as shifted copies stationary kernel. This...
Partial volume effect is an important source of bias in PET images that can be lowered by accounting for the point spread function (PSF) scanner. We measured such a PSF various points clinical scanner and modelled it as product matrices acting image space, taking asymmetrical, shift-varying non-Gaussian character into account (AMP modelling), we integrated this accurate space modelling conventional list-mode OSEM algorithm (EM-AMP reconstruction). showed on one hand when sufficiently high...
The Cosmological Microwave Background (CMB) is of premier importance for the cosmologists to study birth our universe. Unfortunately, most CMB experiments such as COBE, WMAP or Planck do not provide a direct measure cosmological signal; mixed up with galactic foregrounds and point sources. For sake scientific exploitation, measuring requires extracting several different astrophysical components (CMB, Sunyaev-Zel'dovich clusters, dust) form multi-wavelength observations. Mathematically...
The deconvolution of large survey images with millions galaxies requires developing a new generation methods that can take space-variant point spread function into account. These have also to be accurate and fast. We investigate how deep learning might used perform this task. employed U-net neural network architecture learn parameters were adapted for galaxy image processing in supervised setting studied two strategies. first approach is post-processing mere Tikhonov closed-form solution,...
In this article, we describe a new estimate of the Cosmic Microwave Background (CMB) intensity map reconstructed by joint analysis full Planck 2015 data (PR2) and WMAP nine-years. It provides more than mere update CMB introduced in (Bobin et al. 2014b) since it benefits from an improvement component separation method L-GMCA (Local-Generalized Morphological Component Analysis) that allows efficient correlated components 2015). Based on most recent data, further confirm previous results...
We present a new method to estimate shear measurement bias in image simulations that significantly improves the precision with respect current techniques. Our is based on measuring response for individual images. generated sheared versions of same measure how galaxy shape changes small applied shear. This multiplicative each image. In addition, we also measured additive bias. Using noise realizations version allows us compute at very high precision. The estimated sample galaxies then average...
A new dynamic image reconstruction method for PET is proposed. First, a set of exponential temporal basis functions predefined, covering the entire range kinetics (from static through to delta function). Just as in spectral analysis, such selection designed be able handle all possible tissue responses multi-compartmental models. Second, an initial estimate input function defined. The time-dependent radiotracer concentration then modeled (through system matrix algorithm) superposition...
Large-scale anomalies have been reported in CMB data with both WMAP and Planck data. These could be due to foreground residuals or systematic effects, though their confirmation suggests they are not a problem the pipelines. If these fact primordial, then understanding origin is fundamental either validate standard model of cosmology explore new physics. We investigate three other possible issues: 1) trade-off between minimising systematics contamination (with conservative mask) masking, 2)...
4D PET imaging seeks to estimate kinetic parameters of physiological significance through the generation a time series 3D images. Conventionally is reconstructed one frame at time, and then modeling applied as post-reconstruction step desired parameters. Such separated approach does not account for task parameter estimation within reconstruction itself. This work indicates that conventional frame-by-frame maximum likelihood in high noise situations sub-optimal if be performed. As an...
With the increasing number of deep multi-wavelength galaxy surveys, spectral energy distribution (SED) galaxies has become an invaluable tool for studying formation their structures and evolution. In this context, standard analysis relies on simple spectro-photometric selection criteria based a few SED colors. If fully supervised classification already yielded clear achievements, it is not optimal to extract relevant information from data. article, we propose employ very recent advances in...
We propose in this work a framework for synergistic positron emission tomography (PET)/computed (CT) reconstruction using joint generative model as penalty. use penalty function that promotes PET/CT pairs are likely to occur together. The is based on model, namely $\beta$-variational autoencoder ($\beta$-VAE). generates image pair from the same latent variable which contains information shared between two modalities. This sharing of inter-modal can help reduce noise during reconstruction....
The primordial power spectrum describes the initial perturbations in Universe which eventually grew into large-scale structure we observe today, and thereby provides an indirect probe of inflation or other structure-formation mechanisms. Here, introduce a new method to estimate this from empirical cosmic microwave background (CMB) maps. A sparsity-based linear inversion method, coined \textbf{PRISM}, is presented. This technique leverages sparsity prior on features wavelet basis regularise...
In metric theories of gravity with photon number conservation, the luminosity and angular diameter distances are related via Etherington relation, also known as distance-duality relation (DDR). A violation this would rule out standard cosmological paradigm point at presence new physics. We quantify ability Euclid, in combination contemporary surveys, to improve current constraints on deviations from DDR redshift range $0
Deep Learning (DL) has shown remarkable results in solving inverse problems various domains. In particular, the Tikhonet approach is very powerful to deconvolve optical astronomical images (Sureau et al. 2020). Yet, this only uses $\ell_2$ loss, which does not guarantee preservation of physical information (e.g. flux and shape) object reconstructed image. Nammour (2021), a new loss function was proposed framework sparse deconvolution, better preserves shape galaxies reduces pixel error....
ABSTRACT The Euclid mission will observe well over a billion galaxies out to z ∼ 6 and beyond. This offer an unrivalled opportunity investigate several key questions for understanding galaxy formation evolution. first step many of these studies be the selection sample quiescent star-forming galaxies, as is often done in literature by using well-known colour techniques such ‘UVJ’ diagram. However, given limited number filters available telescope, recovery rest-frame colours challenging. We...
The data from the Euclid mission will enable measurement of photometric redshifts, angular positions, and weak lensing shapes for over a billion galaxies. This large dataset allow cosmological analyses using clustering galaxies cosmic shear. cross-correlation (XC) between these probes can tighten constraints it is therefore important to quantify their impact Euclid. In this study we carefully XC not only on final parameter different models, but also nuisance parameters. particular, aim at...
Recovering the Cosmic Microwave Background (CMB) from WMAP data requires galactic foreground emissions to be accurately separated out. Most component separation techniques rely on second order statistics such as Internal Linear Combination (ILC) techniques. In this paper, we present a new 9-year CMB map, with 15 arcmin resolution, which is reconstructed using recently introduced sparse technique, coined Local Generalized Morphological Component Analysis (LGMCA). LGMCA emphasizes sparsity of...
Stage IV weak lensing experiments will offer more than an order of magnitude leap in precision. We must therefore ensure that our analyses remain accurate this new era. Accordingly, previously ignored systematic effects be addressed. In work, we evaluate the impact reduced shear approximation and magnification bias, on information obtained from angular power spectrum. To first-order, statistics shear, a combination convergence, are taken to equal those shear. However, can induce bias...
Clinical use of positron emission tomography (PET) for brain imaging is limited by the partial-volume effect (PVE) induced spatial resolution most scanners. Correction this often performed using a post-reconstruction processing framework involving external information provided an MRI acquisition. This approach has major drawback being very sensitive to unavoidable segmentation and PET-MRI registration mismatches. Under assumption that PVE better compensated when it modeled in reconstruction...