CNRS Research Center for Image Acquisition and Processing for Health
CNRS Research Center for Image Acquisition and Processing for Health
Duchateau N.,CNRS Research Center for Image Acquisition and Processing for Health
IEEE Transactions on Medical Imaging | Year: 2017
Collecting large databases of annotated medical images is crucial for the validation and testing of feature extraction, statistical analysis and machine learning algorithms. Recent advances in cardiac electromechanical modeling and image synthesis provided a framework to generate synthetic images based on realistic mesh simulations. Nonetheless, their potential to augment an existing database with large amounts of synthetic cases requires further investigation. We build upon these works and propose a revised scheme for synthesizing pathological cardiac sequences from real healthy sequences. Our new pipeline notably involves a much easier registration problem to reduce potential artifacts, and takes advantage of mesh correspondences to generate new data from a given case without additional registration. The output sequences are thoroughly examined in terms of quality and usability on a given application: the assessment of myocardial viability, via the generation of 465 synthetic cine MR sequences (15 healthy and 450 with pathological tissue viability [random location, extent and grade, up to myocardial infarct]). We demonstrate that our methodology (i) improves state-of-the-art algorithms in terms of realism and accuracy of the simulated images, and (ii) is well-suited for the generation of large databases at small computational cost. IEEE
Luo J.,Shanghai JiaoTong University |
Zhu Y.,CNRS Research Center for Image Acquisition and Processing for Health
Digital Signal Processing: A Review Journal | Year: 2012
A novel approach for denoising medical images is proposed based on a reconstruction-average mechanism. First, different parts of the original complete spectrum are chosen, from each of which a signal is reconstructed using a singularity function analysis (SFA) model. We finally achieve denoising by averaging these reconstructed signals using the fact that each of them is the sum of the same noise-free signal and an additive noise of varying magnitude. The theoretical ground of such approach is mathematically formulated. The experimental results on both simulated and real monochrome images show that the proposed denoising method allows efficient denoising while maintaining image quality, and presents significant advantages over conventional denoising methods. © 2011 Elsevier Inc.
Langer M.,European Synchrotron Radiation Facility |
Langer M.,CNRS Research Center for Image Acquisition and Processing for Health |
Cloetens P.,European Synchrotron Radiation Facility |
Peyrin F.,European Synchrotron Radiation Facility |
Peyrin F.,CNRS Research Center for Image Acquisition and Processing for Health
IEEE Transactions on Image Processing | Year: 2010
We consider the phase retrieval problem in 3-D holotomography for strongly absorbing objects. Holotomography combines phase retrieval from Fresnel diffraction patterns with tomographic reconstruction to reconstruct the 3-D refractive index distribution. The main interest is the increase in sensitivity of up to three orders of magnitude compared to standard, absorption based tomography. Most existing algorithms are based upon linearization of the forward problem. This is motivated by the large problem size, since it yields computationally efficient solutions. Here, the mixed approach is used, which allows for both strong absorption and long propagation distances. Previous implementations have shown promising results, but in practice often suffer from strong low frequency artifacts. To address this problem, we introduce a homogeneous object assumption through a regularizing term based upon the absorption image. This allows the homogeneous object assumption to be introduced only in the low frequency range. The proportionality constant between absorption and refractive index is assumed to be known. The regularizing parameter is found using the standard L-curve technique. The benefits of our approach are illustrated using data measured at the European Synchrotron Radiation Facility. Low frequency noise in the reconstruction is alleviated, but the result is only quantitative in the areas of the sample where the homogeneous object assumption is fulfilled. © 2010 IEEE.
Bao L.,Harbin Institute of Technology |
Robini M.,CNRS Research Center for Image Acquisition and Processing for Health |
Liu W.,Harbin Institute of Technology |
Liu W.,CNRS Research Center for Image Acquisition and Processing for Health |
Zhu Y.,CNRS Research Center for Image Acquisition and Processing for Health
Medical Image Analysis | Year: 2013
Diffusion tensor magnetic resonance imaging (DT-MRI) is becoming a prospective imaging technique in clinical applications because of its potential for in vivo and non-invasive characterization of tissue organization. However, the acquisition of diffusion-weighted images (DWIs) is often corrupted by noise and artifacts, and the intensity of diffusion-weighted signals is weaker than that of classical magnetic resonance signals. In this paper, we propose a new denoising method for DT-MRI, called structure-adaptive sparse denoising (SASD), which exploits self-similarity in DWIs. We define a similarity measure based on the local mean and on a modified structure-similarity index to find sets of similar patches that are arranged into three-dimensional arrays, and we propose a simple and efficient structure-adaptive window pursuit method to achieve sparse representation of these arrays. The noise component of the resulting structure-adaptive arrays is attenuated by Wiener shrinkage in a transform domain defined by two-dimensional principal component decomposition and Haar transformation. Experiments on both synthetic and real cardiac DT-MRI data show that the proposed SASD algorithm outperforms state-of-the-art methods for denoising images with structural redundancy. Moreover, SASD achieves a good trade-off between image contrast and image smoothness, and our experiments on synthetic data demonstrate that it produces more accurate tensor fields from which biologically relevant metrics can then be computed. © 2013 Elsevier B.V.
Guerin C.,Reanimation Medicale |
Guerin C.,CNRS Research Center for Image Acquisition and Processing for Health
Current Opinion in Critical Care | Year: 2014
PURPOSE OF REVIEW: Prone position can prevent ventilator-induced lung injury in acute respiratory distress syndrome (ARDS) patients receiving conventional mechanical ventilation and, hence, may have the potential to improve survival from this basis. Even though no single randomized controlled trial has proven benefit on patient outcome until recently, two meta-Analyses, one on grouped data and the other on individual data, have shown that patients with PaO2/FIO2 ratio less than 100 mmHg at the time of inclusion did benefit from prone position. As a fifth trial completed recently has shown a significant reduction in mortality in patients with severe and confirmed ARDS from using prone position, the purpose of this review is to revisit prone positioning in ARDS in the light of these new findings. RECENT FINDINGS: In this trial done in patients with severe ARDS severity criteria (PaO2/FIO2 ratio less than 150â€ŠmmHg with positive end expiratory pressure of 5â€ŠcmH2O or more, FIO2 of 60% or more and tidal volume around 6â€Šml/kg predicted body weight) confirmed 12-24â€Šh after the onset of ARDS, the day 28 mortality in the supine group (229 patients) was 32.8 versus 16% in the prone group (237 patients) (Pâ€Š<â€Š0.001). Significant reduction in mortality was confirmed at day 90. SUMMARY: From the combined results of the two meta-Analyses and the last randomized controlled trial, there is a very strong signal to use prone position in patients with severe ARDS, as early as possible and for long sessions. © 2014 Wolters Kluwer Health | Lippincott Williams & Wilkins.
Robini M.C.,CNRS Research Center for Image Acquisition and Processing for Health |
Reissman P.-J.,Amadeus SAS
Journal of Global Optimization | Year: 2013
Simulated annealing (SA) is a generic optimization method that is quite popular because of its ease of implementation and its global convergence properties. However, SA is widely reported to converge very slowly, and it is common practice to allow extra freedom in its design at the expense of losing global convergence guarantees. A natural way to increase the flexibility of SA is to allow the objective function and the communication mechanism to be temperature-dependent, the idea being to gradually reveal the complexity of the optimization problem and to increase the mixing rate at low temperatures. We call this general class of annealing processes stochastic continuation (SC). In the first part of this paper, we introduce SC starting from SA, and we derive simple sufficient conditions for the global convergence of SC. Our main result is interesting in two respects: first, the conditions for global convergence are surprisingly weak - in particular, they do not involve the variations of the objective function with temperature - and second, exponential cooling makes it possible to be arbitrarily close to the best possible convergence speed exponent of SA. The second part is devoted to the application of SC to the problem of producing aesthetically pleasing drawings of undirected graphs. We consider the objective function defined by Kamada and Kawai (Inf Process Lett 31(1):7-15, 1989), which measures the quality of a drawing as a weighted sum of squared differences between Euclidean and graph-theoretic inter-vertex distances. Our experiments show that SC outperforms SA with optimal communication setting both in terms of minimizing the objective function and in terms of standard aesthetic criteria. © 2012 Springer Science+Business Media, LLC.
Lartizien C.,CNRS Research Center for Image Acquisition and Processing for Health |
Aubin J.-B.,University Technologique Of Compiegne |
Buvat I.,University Paris Diderot
IEEE Transactions on Medical Imaging | Year: 2010
Two groups of bootstrap methods have been proposed to estimate the statistical properties of positron emission tomography (PET) images by generating multiple statistically equivalent data sets from few data samples. The first group generates resampled data based on a parametric approach assuming that data from which resampling is performed follows a Poisson distribution while the second group consists of nonparametric approaches. These methods either require a unique original sample or a series of statistically equivalent data that can be list-mode files or sinograms. Previous reports regarding these bootstrap approaches suggest different results. This work compares the accuracy of three of these bootstrap methods for 3-D PET imaging based on simulated data. Two methods are based on a unique file, namely a list-mode based nonparametric (LMNP) method and a sinogram based parametric (SP) method. The third method is a sinogram-based nonparametric (SNP) method. Another original method (extended LMNP) was also investigated, which is an extension of the LMNP methods based on deriving a resampled list-mode file by drawings events from multiple original list-mode files. Our comparison is based on the analysis of the statistical moments estimated on the repeated and resampled data. This includes the probability density function and the moments of order 1 and 2. Results show that the two methods based on multiple original data (SNP and extended LMNP) are the only methods that correctly estimate the statistical parameters. Performances of the LMNP and SP methods are variable. Simulated data used in this study were characterized by a high noise level. Differences among the tested strategies might be reduced with clinical data sets with lower noise. © 2006 IEEE.
Sdika M.,CNRS Research Center for Image Acquisition and Processing for Health
Medical Physics | Year: 2015
Purpose: To present a method to enrich atlases for atlas based segmentation. Such enriched atlases can then be used as a single atlas or within a multiatlas framework. Methods: In this paper, machine learning techniques have been used to enhance the atlas based segmentation approach. The enhanced atlas defined in this work is a pair composed of a gray level image alongside an image of multiclass classifiers with one classifier per voxel. Each classifier embeds local information from the whole training dataset that allows for the correction of some systematic errors in the segmentation and accounts for the possible local registration errors. The authors also propose to use these images of classifiers within a multiatlas framework: results produced by a set of such local classifier atlases can be combined using a label fusion method. Results: Experiments have been made on the in vivo images of the IBSR dataset and a comparison has been made with several state-of-the-art methods such as FreeSurfer and the multiatlas nonlocal patch based method of Coupé or Rousseau. These experiments show that their method is competitive with state-of-the-art methods while having a low computational cost. Further enhancement has also been obtained with a multiatlas version of their method. It is also shown that, in this case, nonlocal fusion is unnecessary. The multiatlas fusion can therefore be done efficiently. Conclusions: The single atlas version has similar quality as state-of-the-arts multiatlas methods but with the computational cost of a naive single atlas segmentation. The multiatlas version offers a improvement in quality and can be done efficiently without a nonlocal strategy. © 2015 American Association of Physicists in Medicine.
Schaerer J.,CNRS Research Center for Image Acquisition and Processing for Health |
Casta C.,CNRS Research Center for Image Acquisition and Processing for Health |
Pousin J.,CNRS Research Center for Image Acquisition and Processing for Health |
Clarysse P.,CNRS Research Center for Image Acquisition and Processing for Health
Medical Image Analysis | Year: 2010
Strong prior models are a prerequisite for reliable spatio-temporal cardiac image analysis. While several cardiac models have been presented in the past, many of them are either too complex for their parameters to be estimated on the sole basis of MR Images, or overly simplified. In this paper, we present a novel dynamic model, based on the equation of dynamics for elastic materials and on Fourier filtering. The explicit use of dynamics allows us to enforce periodicity and temporal smoothness constraints. We propose an algorithm to solve the continuous dynamical problem associated to numerically adapting the model to the image sequence. Using a simple 1D example, we show how temporal filtering can help removing noise while ensuring the periodicity and smoothness of solutions. The proposed dynamic model is quantitatively evaluated on a database of 15 patients which shows its performance and limitations. Also, the ability of the model to capture cardiac motion is demonstrated on synthetic cardiac sequences. Moreover, existence, uniqueness of the solution and numerical convergence of the algorithm can be demonstrated. © 2010 Elsevier B.V.
Liebgott H.,CNRS Research Center for Image Acquisition and Processing for Health
Proceedings - IEEE Ultrasonics Symposium | Year: 2010
Transverse Oscillations (TO) is the name given to a technique able toproduce ultrasound images with oscillations in both axial and lateraldirections. Such images have shown their potential for improving motionestimation. In this paper we present an alternative to Time Domain TransverseOscillations (TDTO) beamforming. Conventional approaches to TO use Fraunhofferapproximation to design their beamformer, which is not respected for broadbandsignals. We propose to proceed to a Fourier Transform of the raw Rf signalsbefore apodization in order to separate the contribution of each frequency bandand adapt the apodization function to each frequency. This method is refered toas Fourier Domain Transverse Oscillations (FDTO) beamforming. FDTO is validatedby simulations on PSFs obtained with FDTO and TDTO. Their comparison showshigher normalized cross-correlation with FDTO than TDTO for expected transverseoscillations wavelength between 1 and 2.5 mm in our configuration. FDTO is alsovalidated on a speckle phantom. Visually FDTO PSFs, PSFs and specke images 2Dspectra have lost the diagonal orientation of their pattern. © 2010 IEEE.