Laboratoire des Systèmes Perceptifs

We report the first experimental demonstration of time-reversal focusing with electromagnetic waves. An antenna transmits a 1-micros electromagnetic pulse at a central frequency of 2.45 GHz in a high-Q cavity. Another antenna records the strongly reverberated signal. The time-reversed wave is built and transmitted back by the same antenna acting now as a time-reversal mirror. The wave is found to converge to its initial source and is compressed in time. The quality of focusing is determined by the frequency bandwidth and the spectral correlations of the field within the cavity.

Developing minimally invasive brain surgery by high-intensity focused ultrasound beams is of great interest in cancer therapy. However, the skull induces strong aberrations both in phase and amplitude, resulting in a severe degradation of the beam shape. Thus, an efficient brain tumor therapy would require an adaptive focusing, taking into account the effects of the skull. In this paper, we will show that the acoustic properties of the skull can be deduced from high resolution CT scans and used to achieve a noninvasive adaptive focusing. Simulations have been performed with a full 3-D finite differences code, taking into account all the heterogeneities inside the skull. The set of signals to be emitted in order to focus through the skull can thus be computed. The complete adaptive focusing procedure based on prior CT scans has been experimentally validated. This could have promising applications in brain tumor hyperthermia but also in transcranial ultrasonic imaging.

Abstract In order to determine surface heat flux at large scale from space measurement, it is necessary to introduce a bulk temperature for the whole pixel area for heterogeneous and non isothermal surfaces. This temperature should be measurable from space and should be related to the corresponding fluxes. A possible definition of such a surface temperature is discussed and its relationship with the radiation and sensible heat fluxes is given in the first part of this paper. This temperature, referred to as the radiometric temperature 〈Tsr〉, depends not only on the distributions of surface temperature and emissivity within a pixel but also on the channel used to measure it. This dependence is modeled and it is shown that if the surface emissivity and surface temperature variations are small within a pixel, 〈Tsr 〉 is equivalent to the average of surface temperatures within this pixel. In some cases, it is possible to get the average and the variance of the distribution of the surface temperatures within a pixel from the measurement of 〈Tsr 〉 in at least two channels. In order to measure this radiometric temperature from space, it is necessary to separate surface temperature and surface emissivity from the observed radiance. We review several approaches in the second part of this paper and present an improved TISI (Temperature‐Independent Spectral Indices) for deriving emis‐sivities from AVHRR data. Once the emissivities are measured, the radiometric temperature may be derived using local split‐window algorithms. We propose a new split window algorithm which takes into account both the spectral emissivities of the surface and the corrections for large variations of the atmospheric water vapor content. Since it is very difficult to obtain in‐situ surface temperature relevant for AVHRR pixel, we just give in the last part of this paper an estimation of the errors of models by comparing, over different sites and during different seasons, the surface temperature retrieved from this local split window algorithm and from several published split window algorithms for AVHRR/2 data using in all the cases the values of emissivjties derived from the method discussed in the second part. The results of this comparison show that algorithms derived by Prata and Platt (1991), Sobrino et al. (1991, 1993), Ulivieri et al. (1994) and the proposed local split window give comparable results, while a former algorithm of Becker and Li (1990b) gives a systematic surface temperature over‐estimation of about 1.5 K with respect to the most recent methods, and other algorithms (Price, 1984; Ulivieri et al. 1985; and NESDIS) can be significantly different.

Magnetic resonance elastography (MRE) is a non-invasive imaging technique used to visualise and quantify mechanical properties of tissue, providing information beyond what can be currently achieved with standard MR sequences and could, for instance, provide new insight into pathological processes in the brain. This study uses the MRE technique at 3 T to extract the complex shear modulus for in vivo brain tissue utilizing a full three-dimensional approach to reconstruction, removing contributions of the dilatational wave by application of the curl operator. A calibrated phantom is used to benchmark the MRE measurements, and in vivo results are presented for healthy volunteers. The results provide data for in vivo brain storage modulus (G'), finding grey matter (3.1 kPa) to be significantly stiffer than white matter (2.7 kPa). The first in vivo loss modulus (G'') measurements show no significant difference between grey matter (2.5 kPa) and white matter (2.5 kPa).

This paper describes a new technique for two-dimensional (2-D) imaging of the motion vector at a very high frame rate with ultrasound. Its potential is experimentally demonstrated for transient elastography. But, beyond this application, it also could be promising for color flow and reflectivity imaging. To date, only axial displacements induced in human tissues by low-frequency vibrators were measured during transient elastography. The proposed technique allows us to follow both axial and lateral displacements during the shear wave propagation and thus should improve Young's modulus image reconstruction. The process is a combination of several ideas well-known in ultrasonic imaging: ultra-fast imaging, multisynthetic aperture beamforming, 1-D speckle tracking, and compound imaging. Classical beamforming in the transmit mode is replaced here by a single plane wave insonification increasing the frame rate by at least a factor of 128. The beamforming is achieved only in the receive mode on two independent subapertures. Comparison of successive frames by a classical 1-D speckle tracking algorithm allows estimation of displacements along two different directions linked to the subapertures beams. The variance of the estimates is finally improved by tilting the emitting plane wave at each insonification, thus allowing reception of successive decorrelated speckle patterns.

Two main questions are at the center of this paper. The first one concerns the choice of a rheological model in the frequency range of transient elastography, sonoelasticity or NMR elastography for soft solids (20-1000 Hz). Transient elastography experiments based on plane shear waves that propagate in an Agar-gelatin phantom or in bovine muscles enable one to quantify their viscoelastic properties. The comparison of these experimental results to the prediction of the two simplest rheological models indicate clearly that Voigt's model is the better. The second question studied in the paper deals with the feasibility of quantitative viscosity mapping using inverse problem algorithm. In the ideal situation where plane shear waves propagate in a sample, a simple inverse problem based on the Helmholtz equation correctly retrieves both elasticity and viscosity. In a more realistic situation with nonplane shear waves, this simple approach fails. Nevertheless, it is shown that quantitative viscosity mapping is still possible if one uses an appropriate inverse problem that fully takes into account diffraction in solids.

MR-elastography is a new technique for assessing the viscoelastic properties of tissue. One current focus of elastography is the provision of new physical parameters for improving the specificity in breast cancer diagnosis. This analysis describes a technique to extend the reconstruction to anisotropic elastic properties in terms of a so-called transversely isotropic model. Viscosity is treated as being isotropic. The particular model chosen for the anisotropy is appealing because it is capable of describing elastic shear anisotropy of parallel fibers. The dependence of the reconstruction on the particular choice of Poisson's ratio is eliminated by extracting the compressional displacement contribution using the Helmholtz-Hodge decomposition. Results are presented for simulations, a polyvinyl alcohol breast phantom, excised beef muscle, and measurements in two patients with breast lesions (invasive ductal carcinoma and fibroadenoma). The results show enhanced anisotropic and viscous properties inside the lesions and an indication for preferred fiber orientation.

For pt.II see ibid., vol.39, no.5, p.567-78 (1992). A theoretical model for time-reversal cavities to optimize focusing in homogeneous and inhomogeneous media is described. The concept of the cavity can be understood as the most realistic approximation to an exact three-dimensional (3-D) time-reversal of ultrasonic fields; it is also a generalization of the time-reversal mirrors realized experimentally in the laboratory. The proposed method is based on an approach in the transient regime that is more general than the monochromatic formalism used in optics to analyze the phase conjugation mirrors efficiency. This method uses impulse diffraction theory to obtain the impulse response of the cavity in any inhomogeneous medium. An original interpretation of the limitations due to diffraction observed in wave field propagation in terms of the different waves generated inside the cavity is also proposed. The time-reversal focusing process using a closed cavity in a weakly inhomogeneous medium is compared with more classical techniques to compensate wavefront distortions, thus illustrating the focusing improvement due to the time-reversal method.

A detailed chemical kinetic model has been used to study dimethyl ether (DME) oxidation over a wide range of conditions. Experimental results obtained in a jet-stirred reactor (JSR) at 1 and 10 atm, 0.2≤ϕ≤2.5, and 800≤T≤1300 K were modeled, in addition to those generated in a shock tube at 13 and 40 bar, ϕ=1.0 and 650≤T≤1300 K. The JSR results are particularly valuable as they include concentration profiles of reactants, intermediates, and products pertinent to the oxidation of DME. These data test the kinetic model severely, as it must be able to predict the correct distribution and concentrations of intermediate and final products formed in the oxidation process. Additionally, the shock-tube results are very useful, as they were taken at low temperatures and at high pressures, and thus undergo negative temperature dependence (NTC) behavior. This behavior is characteristic of the oxidation of saturated hydrocarbon fuels, (e.g., the primary reference fuels, n-heptane and iso-octane) under similar conditions. The numerical model consists of 78 chemical species and 336 chemical reactions. The thermodynamic properties of unknown species pertaining to DME oxidation were calculated using THERM. © 1998 John Wiley & Sons, Inc. Int J Chem Kinet 30: 229–241, 1998.

Considerable progress has recently been made in natural language processing: deep learning algorithms are increasingly able to generate, summarize, translate and classify texts. Yet, these language models still fail to match the language abilities of humans. Predictive coding theory offers a tentative explanation to this discrepancy: while language models are optimized to predict nearby words, the human brain would continuously predict a hierarchy of representations that spans multiple timescales. To test this hypothesis, we analysed the functional magnetic resonance imaging brain signals of 304 participants listening to short stories. First, we confirmed that the activations of modern language models linearly map onto the brain responses to speech. Second, we showed that enhancing these algorithms with predictions that span multiple timescales improves this brain mapping. Finally, we showed that these predictions are organized hierarchically: frontoparietal cortices predict higher-level, longer-range and more contextual representations than temporal cortices. Overall, these results strengthen the role of hierarchical predictive coding in language processing and illustrate how the synergy between neuroscience and artificial intelligence can unravel the computational bases of human cognition.

To focus ultrasonic waves in an unknown inhomogeneous medium using a phased array, one has to calculate the optimal set of signals to be applied on the transducers of the array. In the case of time-reversal mirrors, one assumes that a source is available at the focus, providing the Green's function of this point. In this paper, the robustness of this time-reversal method is investigated when loss of information breaks the time-reversal invariance. It arises in dissipative media or when the field radiated by the source is not entirely measured by the limited aperture of a time-reversal mirror. However, in both cases, linearity and reciprocity relations ensure time reversal to achieve a spatiotemporal matched filtering. Nevertheless, though it provides robustness to this method, no constraints are imposed on the field out of the focus and sidelobes may appear. Another approach consists of measuring the Green's functions associated to the focus but also to neighboring points. Thus, the whole information characterizing the medium is known and the inverse source problem can be solved. A matrix formalism of the propagation operator is introduced to compare the time-reversal and inverse filter techniques. Moreover, experiments investigated in various media are presented to illustrate this comparison.

A classical theorem of statistical optics, the van Cittert–Zernike theorem, is generalized to pulse echo ultrasound. This theorem fully describes the second-order statistics of the spatial fluctuations (the spatial covariance) of the field produced by an incoherent source. As a random scattering medium is insonified, it behaves as an incoherent source. The van Cittert–Zernike theorem can thus predict the spatial covariance of the pressure field backscattered by a random medium. It is shown that this spatial covariance and the incident energy diagram are Fourier pairs. In the case of a focused illumination, the spatial covariance of the backscattered pressure field is proportional to the autocorrelation of the transmitting aperture function. This is independent of frequency and of F/ number. Experimental results obtained with a linear array are in good agreement with theoretical expectations. The implications of this theorem in speckle reduction and in focusing in nonhomogenous media are discussed.

In recent years, time-reversal (TR) mirrors have been developed that create TR waves for ultrasonic transient fields propagating through complex media. A TR wave back propagates and refocuses exactly at its initial source. However, because of diffraction, even if the source is pointlike the wave refocuses on a spot size that cannot be smaller than half a wavelength. Here, by using a TR interpretation of this limit, we show that this latter limitation can be overcome if the source is replaced by its TR image. This new device acts as an acoustic sink that absorbs the TR wave. Here we report the first experimental result obtained with an acoustic sink where a focal spot size of less than 1/14th of one wavelength is recorded.

Kinetic reaction mechanisms are necessary for modeling the combustion, oxidation and ignition of commercial fuels consisting of complex mixtures of hydrocarbons. Since they are generally too complex to be considered in the models directly, simple model-fuels are preferred. These model-fuels consist in a simple mixture of hydrocarbons for which kinetic oxidation models are validated. The oxidation of a large variety of hydrocarbons was studied experimentally in a jet-stirred reactor to build the needed kinetic reaction mechanisms. These detailed kinetic reaction mechanisms were assembled to model the oxidation of commercial fuels. The capabilities of these kinetic models to simulate the oxidation of natural gas, kerosene and gas oil are presented together with needs for new kinetic measurements.

International audience

An original formalism that allows a thorough understanding of cross-correlation-based phase aberration correction techniques in a scattering medium is developed. This formalism is based on the analysis of the second-order statistics of the pressure field scattered by a random distribution of scatterers. One of the major interests of this analysis is the ability it provides to evaluate and monitor the convergence of phase aberration correction techniques. This is achieved by the computation of a single parameter C that serves as a focusing criterion. The theoretical developments are validated experimentally.

The time-reversal process is applied to focus pulsed ultrasonic waves through the human skull bone. The aim here is to treat brain tumors, which are difficult to reach with classical surgery means. Such a surgical application requires precise control of the size and location of the therapeutic focal beam. The severe ultrasonic attenuation in the skull reduces the efficiency of the time reversal process. Nevertheless, an improvement of the time reversal process in absorbing media has been investigated and applied to the focusing through the skull [J.-L. Thomas and M. Fink, IEEE Trans. Ultrason. Ferroelectr. Freq. Control 43, 1122-1129 (1996)]. Here an extension of this technique is presented in order to focus on a set of points surrounding an initial artificial source implanted in the tissue volume to treat. From the knowledge of the Green's function matched to this initial source location a new Green's function matched to various points of interest is deduced in order to treat the whole volume. In a homogeneous medium, conventional steering consists of tilting the wave front focused on the acoustical source. In a heterogeneous medium, this process is only valid for small angles or when aberrations are located in a layer close to the array. It is shown here how to extend this method to aberrating and absorbing layers, like the skull bone, located at any distance from the array of transducers.

We present an experimental demonstration showing that, contrary to first intuition, the more scattering a mesoscopic medium is, the more information can be conveyed through it. We used a multiple input-multiple output configuration: a multichannel ultrasonic time-reversal antenna is used to transmit random series of bits simultaneously to different receivers which were only a few wavelengths apart. Whereas the transmission is free of error when multiple scattering occurs in the propagation medium, the error rate is huge in a homogeneous medium.

The relation between a stimulus and the evoked brain response can shed light on perceptual processes within the brain. Signals derived from this relation can also be harnessed to control external devices for Brain Computer Interface (BCI) applications. While the classic event-related potential (ERP) is appropriate for isolated stimuli, more sophisticated "decoding" strategies are needed to address continuous stimuli such as speech, music or environmental sounds. Here we describe an approach based on Canonical Correlation Analysis (CCA) that finds the optimal transform to apply to both the stimulus and the response to reveal correlations between the two. Compared to prior methods based on forward or backward models for stimulus-response mapping, CCA finds significantly higher correlation scores, thus providing increased sensitivity to relatively small effects, and supports classifier schemes that yield higher classification scores. CCA strips the brain response of variance unrelated to the stimulus, and the stimulus representation of variance that does not affect the response, and thus improves observations of the relation between stimulus and response.

A non-invasive protocol for transcranial brain tissue ablation with ultrasound is studied and validated in vitro. The skull induces strong aberrations both in phase and in amplitude, resulting in a severe degradation of the beam shape. Adaptive corrections of the distortions induced by the skull bone are performed using a previous 3D computational tomography scan acquisition (CT) of the skull bone structure. These CT scan data are used as entry parameters in a FDTD (finite differences time domain) simulation of the full wave propagation equation. A numerical computation is used to deduce the impulse response relating the targeted location and the ultrasound therapeutic array, thus providing a virtual time-reversal mirror. This impulse response is then time-reversed and transmitted experimentally by a therapeutic array positioned exactly in the same referential frame as the one used during CT scan acquisitions. In vitro experiments are conducted on monkey and human skull specimens using an array of 300 transmit elements working at a central frequency of 1 MHz. These experiments show a precise refocusing of the ultrasonic beam at the targeted location with a positioning error lower than 0.7 mm. The complete validation of this transcranial adaptive focusing procedure paves the way to in vivo animal and human transcranial HIFU investigations.