Estimation of Symmetric Channels for Discrete Cosine Transform Type-I Multicarrier Systems: A Compressed Sensing Approach

The problem of channel estimation for multicarrier communications is addressed. We focus on systems employing the Discrete Cosine Transform Type-I (DCT1) even at both the transmitter and the receiver, presenting an algorithm which achieves an accurate estimation of symmetric channel filters using only a small number of training symbols. The solution is obtained by using either matrix inversion or compressed sensing algorithms. We provide the theoretical results which guarantee the validity of the proposed technique for the DCT1. Numerical simulations illustrate the good behaviour of the proposed algorithm.

Development of cardiovascular magnetic resonance imaging techniques for improved characterization of atherosclerotic disease

Atherosclerosis is a chronic cardiovascular disease that involves the thicken¬ing of the artery walls as well as the formation of plaques (lesions) causing the narrowing of the lumens, in vessels such as the aorta, the coronary and the carotid arteries. Magnetic resonance imaging (MRI) is a promising modality for the assessment of atherosclerosis, as it is a non-invasive and patient-friendly procedure that does not use ionizing radiation. MRI offers high soft tissue con¬trast already without the need of intravenous contrast media; while modifica¬tion of the MR pulse sequences allows for further adjustment of the contrast for specific diagnostic needs. As such, MRI can create angiographic images of the vessel lumens to assess stenoses at the late stage of the disease, as well as blood flow-suppressed images for the early investigation of the vessel wall and the characterization of the atherosclerotic plaques. However, despite the great technical progress that occurred over the past two decades, MRI is intrinsically a low sensitive technique and some limitations still exist in terms of accuracy and performance. A major challenge for coronary artery imaging is respiratory motion. State- of-the-art diaphragmatic navigators rely on an indirect measure of motion, per¬form a ID correction, and have long and unpredictable scan time. In response, self-navigation (SM) strategies have recently been introduced that offer 100% scan efficiency and increased ease of use. SN detects respiratory motion di¬rectly from the image data obtained at the level of the heart, and retrospectively corrects the same data before final image reconstruction. Thus, SN holds po-tential for multi-dimensional motion compensation. To this regard, this thesis presents novel SN methods that estimate 2D and 3D motion parameters from aliased sub-images that are obtained from the same raw data composing the final image. Combination of all corrected sub-images produces a final image with reduced motion artifacts for the visualization of the coronaries. The first study (section 2.2, 2D Self-Navigation with Compressed Sensing) consists of a method for 2D translational motion compensation. Here, the use of com- pressed sensing (CS) reconstruction is proposed and investigated to support motion detection by reducing aliasing artifacts. In healthy human subjects, CS demonstrated an improvement in motion detection accuracy with simula¬tions on in vivo data, while improved coronary artery visualization was demon¬strated on in vivo free-breathing acquisitions. However, the motion of the heart induced by respiration has been shown to occur in three dimensions and to be more complex than a simple translation. Therefore, the second study (section 2.3,3D Self-Navigation) consists of a method for 3D affine motion correction rather than 2D only. Here, different techniques were adopted to reduce background signal contribution in respiratory motion tracking, as this can be adversely affected by the static tissue that surrounds the heart. The proposed method demonstrated to improve conspicuity and vi¬sualization of coronary arteries in healthy and cardiovascular disease patient cohorts in comparison to a conventional ID SN method. In the third study (section 2.4, 3D Self-Navigation with Compressed Sensing), the same tracking methods were used to obtain sub-images sorted according to the respiratory position. Then, instead of motion correction, a compressed sensing reconstruction was performed on all sorted sub-image data. This process ex¬ploits the consistency of the sorted data to reduce aliasing artifacts such that the sub-image corresponding to the end-expiratory phase can directly be used to visualize the coronaries. In a healthy volunteer cohort, this strategy improved conspicuity and visualization of the coronary arteries when compared to a con¬ventional ID SN method. For the visualization of the vessel wall and atherosclerotic plaques, the state- of-the-art dual inversion recovery (DIR) technique is able to suppress the signal coming from flowing blood and provide positive wall-lumen contrast. How¬ever, optimal contrast may be difficult to obtain and is subject to RR variability. Furthermore, DIR imaging is time-inefficient and multislice acquisitions may lead to prolonged scanning times. In response and as a fourth study of this thesis (chapter 3, Vessel Wall MRI of the Carotid Arteries), a phase-sensitive DIR method has been implemented and tested in the carotid arteries of a healthy volunteer cohort. By exploiting the phase information of images acquired after DIR, the proposed phase-sensitive method enhances wall-lumen contrast while widens the window of opportunity for image acquisition. As a result, a 3-fold increase in volumetric coverage is obtained at no extra cost in scanning time, while image quality is improved. In conclusion, this thesis presented novel methods to address some of the main challenges for MRI of atherosclerosis: the suppression of motion and flow artifacts for improved visualization of vessel lumens, walls and plaques. Such methods showed to significantly improve image quality in human healthy sub¬jects, as well as scan efficiency and ease-of-use of MRI. Extensive validation is now warranted in patient populations to ascertain their diagnostic perfor¬mance. Eventually, these methods may bring the use of atherosclerosis MRI closer to the clinical practice. Résumé L'athérosclérose est une maladie cardiovasculaire chronique qui implique le épaississement de la paroi des artères, ainsi que la formation de plaques (lé¬sions) provoquant le rétrécissement des lumières, dans des vaisseaux tels que l'aorte, les coronaires et les artères carotides. L'imagerie par résonance magné¬tique (IRM) est une modalité prometteuse pour l'évaluation de l'athérosclérose, car il s'agit d'une procédure non-invasive et conviviale pour les patients, qui n'utilise pas des rayonnements ionisants. L'IRM offre un contraste des tissus mous très élevé sans avoir besoin de médias de contraste intraveineux, tan¬dis que la modification des séquences d'impulsions de RM permet en outre le réglage du contraste pour des besoins diagnostiques spécifiques. À ce titre, l'IRM peut créer des images angiographiques des lumières des vaisseaux pour évaluer les sténoses à la fin du stade de la maladie, ainsi que des images avec suppression du flux sanguin pour une première enquête des parois des vais¬seaux et une caractérisation des plaques d'athérosclérose. Cependant, malgré les grands progrès techniques qui ont eu lieu au cours des deux dernières dé¬cennies, l'IRM est une technique peu sensible et certaines limitations existent encore en termes de précision et de performance. Un des principaux défis pour l'imagerie de l'artère coronaire est le mou¬vement respiratoire. Les navigateurs diaphragmatiques de pointe comptent sur une mesure indirecte de mouvement, effectuent une correction 1D, et ont un temps d'acquisition long et imprévisible. En réponse, les stratégies d'auto- navigation (self-navigation: SN) ont été introduites récemment et offrent 100% d'efficacité d'acquisition et une meilleure facilité d'utilisation. Les SN détectent le mouvement respiratoire directement à partir des données brutes de l'image obtenue au niveau du coeur, et rétrospectivement corrigent ces mêmes données avant la reconstruction finale de l'image. Ainsi, les SN détiennent un poten¬tiel pour une compensation multidimensionnelle du mouvement. A cet égard, cette thèse présente de nouvelles méthodes SN qui estiment les paramètres de mouvement 2D et 3D à partir de sous-images qui sont obtenues à partir des mêmes données brutes qui composent l'image finale. La combinaison de toutes les sous-images corrigées produit une image finale pour la visualisation des coronaires ou les artefacts du mouvement sont réduits. La première étude (section 2.2,2D Self-Navigation with Compressed Sensing) traite d'une méthode pour une compensation 2D de mouvement de translation. Ici, on étudie l'utilisation de la reconstruction d'acquisition comprimée (compressed sensing: CS) pour soutenir la détection de mouvement en réduisant les artefacts de sous-échantillonnage. Chez des sujets humains sains, CS a démontré une amélioration de la précision de la détection de mouvement avec des simula¬tions sur des données in vivo, tandis que la visualisation de l'artère coronaire sur des acquisitions de respiration libre in vivo a aussi été améliorée. Pourtant, le mouvement du coeur induite par la respiration se produit en trois dimensions et il est plus complexe qu'un simple déplacement. Par conséquent, la deuxième étude (section 2.3, 3D Self-Navigation) traite d'une méthode de cor¬rection du mouvement 3D plutôt que 2D uniquement. Ici, différentes tech¬niques ont été adoptées pour réduire la contribution du signal du fond dans le suivi de mouvement respiratoire, qui peut être influencé négativement par le tissu statique qui entoure le coeur. La méthode proposée a démontré une amélioration, par rapport à la procédure classique SN de correction 1D, de la visualisation des artères coronaires dans le groupe de sujets sains et des pa¬tients avec maladies cardio-vasculaires. Dans la troisième étude (section 2.4,3D Self-Navigation with Compressed Sensing), les mêmes méthodes de suivi ont été utilisées pour obtenir des sous-images triées selon la position respiratoire. Au lieu de la correction du mouvement, une reconstruction de CS a été réalisée sur toutes les sous-images triées. Cette procédure exploite la cohérence des données pour réduire les artefacts de sous- échantillonnage de telle sorte que la sous-image correspondant à la phase de fin d'expiration peut directement être utilisée pour visualiser les coronaires. Dans un échantillon de volontaires en bonne santé, cette stratégie a amélioré la netteté et la visualisation des artères coronaires par rapport à une méthode classique SN ID. Pour la visualisation des parois des vaisseaux et de plaques d'athérosclérose, la technique de pointe avec double récupération d'inversion (DIR) est capa¬ble de supprimer le signal provenant du sang et de fournir un contraste posi¬tif entre la paroi et la lumière. Pourtant, il est difficile d'obtenir un contraste optimal car cela est soumis à la variabilité du rythme cardiaque. Par ailleurs, l'imagerie DIR est inefficace du point de vue du temps et les acquisitions "mul- tislice" peuvent conduire à des temps de scan prolongés. En réponse à ce prob¬lème et comme quatrième étude de cette thèse (chapitre 3, Vessel Wall MRI of the Carotid Arteries), une méthode de DIR phase-sensitive a été implémenté et testé

Ricostruzione di immagini in norma L1

In questa tesi è stato trattato il problema della ricostruzione di immagini di tomografia computerizzata considerando un modello che utilizza la variazione totale come termine di regolarizzazione e la norma 1 come fidelity term (modello TV/L1). Il problema è stato risolto modificando un metodo di minimo alternato utilizzato per il deblurring e denoising di immagini affette da rumore puntuale. Il metodo è stato testato nel caso di rumore gaussiano e geometria fan beam e parallel beam. Infine vengono riportati i risultati ottenuti dalle sperimentazioni.

Inversione della trasformata di Laplace mediante regolarizzazione con norma L1

I dati derivanti da spettroscopia NMR sono l'effetto di fenomeni descritti attraverso la trasformata di Laplace della sorgente che li ha prodotti. Ci si riferisce a un problema inverso con dati discreti ed in relazione ad essi nasce l'esigenza di realizzare metodi numerici per l'inversione della trasformata di Laplace con dati discreti che è notoriamente un problema mal posto e pertanto occorre ricorrere a metodi di regolarizzazione. In questo contesto si propone una variante ai modelli presenti il letteratura che fanno utilizzo della norma L2, introducendo la norma L1.

Un metodo di tipo Newton per la ricostruzione di immagini con norma L1

Scopo dell'opera è implementare in maniera efficiente ed affidabile un metodo di tipo Newton per la ricostruzione di immagini con termine regolativo in norma L1. In particolare due metodi, battezzati "OWL-QN per inversione" e "OWL-QN precondizionato", sono presentati e provati con numerose sperimentazioni. I metodi sono generati considerando le peculiarità del problema e le proprietà della trasformata discreta di Fourier. I risultati degli esperimenti numerici effettuati mostrano la bontà del contributo proposto, dimostrando la loro superiorità rispetto al metodo OWL-QN presente in letteratura, seppure adattato alle immagini.

Infrared image enhancement based on hybrid-domain consideration and data fusion methods

Au cours des dernières décennies, l’effort sur les applications de capteurs infrarouges a largement progressé dans le monde. Mais, une certaine difficulté demeure, en ce qui concerne le fait que les objets ne sont pas assez clairs ou ne peuvent pas toujours être distingués facilement dans l’image obtenue pour la scène observée. L’amélioration de l’image infrarouge a joué un rôle important dans le développement de technologies de la vision infrarouge de l’ordinateur, le traitement de l’image et les essais non destructifs, etc. Cette thèse traite de la question des techniques d’amélioration de l’image infrarouge en deux aspects, y compris le traitement d’une seule image infrarouge dans le domaine hybride espacefréquence, et la fusion d’images infrarouges et visibles employant la technique du nonsubsampled Contourlet transformer (NSCT). La fusion d’images peut être considérée comme étant la poursuite de l’exploration du modèle d’amélioration de l’image unique infrarouge, alors qu’il combine les images infrarouges et visibles en une seule image pour représenter et améliorer toutes les informations utiles et les caractéristiques des images sources, car une seule image ne pouvait contenir tous les renseignements pertinents ou disponibles en raison de restrictions découlant de tout capteur unique de l’imagerie. Nous examinons et faisons une enquête concernant le développement de techniques d’amélioration d’images infrarouges, et ensuite nous nous consacrons à l’amélioration de l’image unique infrarouge, et nous proposons un schéma d’amélioration de domaine hybride avec une méthode d’évaluation floue de seuil amélioré, qui permet d’obtenir une qualité d’image supérieure et améliore la perception visuelle humaine. Les techniques de fusion d’images infrarouges et visibles sont établies à l’aide de la mise en oeuvre d’une mise en registre précise des images sources acquises par différents capteurs. L’algorithme SURF-RANSAC est appliqué pour la mise en registre tout au long des travaux de recherche, ce qui conduit à des images mises en registre de façon très précise et des bénéfices accrus pour le traitement de fusion. Pour les questions de fusion d’images infrarouges et visibles, une série d’approches avancées et efficaces sont proposés. Une méthode standard de fusion à base de NSCT multi-canal est présente comme référence pour les approches de fusion proposées suivantes. Une approche conjointe de fusion, impliquant l’Adaptive-Gaussian NSCT et la transformée en ondelettes (Wavelet Transform, WT) est propose, ce qui conduit à des résultats de fusion qui sont meilleurs que ceux obtenus avec les méthodes non-adaptatives générales. Une approche de fusion basée sur le NSCT employant la détection comprime (CS, compressed sensing) et de la variation totale (TV) à des coefficients d’échantillons clairsemés et effectuant la reconstruction de coefficients fusionnés de façon précise est proposée, qui obtient de bien meilleurs résultats de fusion par le biais d’une pré-amélioration de l’image infrarouge et en diminuant les informations redondantes des coefficients de fusion. Une procédure de fusion basée sur le NSCT utilisant une technique de détection rapide de rétrécissement itératif comprimé (fast iterative-shrinking compressed sensing, FISCS) est proposée pour compresser les coefficients décomposés et reconstruire les coefficients fusionnés dans le processus de fusion, qui conduit à de meilleurs résultats plus rapidement et d’une manière efficace.

Transmission electron tomography: quality assessment and enhancement for three-dimensional imaging of nanostructures

Nanotechnology has revolutionised humanity's capability in building microscopic systems by manipulating materials on a molecular and atomic scale. Nan-osystems are becoming increasingly smaller and more complex from the chemical perspective which increases the demand for microscopic characterisation techniques. Among others, transmission electron microscopy (TEM) is an indispensable tool that is increasingly used to study the structures of nanosystems down to the molecular and atomic scale. However, despite the effectivity of this tool, it can only provide 2-dimensional projection (shadow) images of the 3D structure, leaving the 3-dimensional information hidden which can lead to incomplete or erroneous characterization. One very promising inspection method is Electron Tomography (ET), which is rapidly becoming an important tool to explore the 3D nano-world. ET provides (sub-)nanometer resolution in all three dimensions of the sample under investigation. However, the fidelity of the ET tomogram that is achieved by current ET reconstruction procedures remains a major challenge. This thesis addresses the assessment and advancement of electron tomographic methods to enable high-fidelity three-dimensional investigations. A quality assessment investigation was conducted to provide a quality quantitative analysis of the main established ET reconstruction algorithms and to study the influence of the experimental conditions on the quality of the reconstructed ET tomogram. Regular shaped nanoparticles were used as a ground-truth for this study. It is concluded that the fidelity of the post-reconstruction quantitative analysis and segmentation is limited, mainly by the fidelity of the reconstructed ET tomogram. This motivates the development of an improved tomographic reconstruction process. In this thesis, a novel ET method was proposed, named dictionary learning electron tomography (DLET). DLET is based on the recent mathematical theorem of compressed sensing (CS) which employs the sparsity of ET tomograms to enable accurate reconstruction from undersampled (S)TEM tilt series. DLET learns the sparsifying transform (dictionary) in an adaptive way and reconstructs the tomogram simultaneously from highly undersampled tilt series. In this method, the sparsity is applied on overlapping image patches favouring local structures. Furthermore, the dictionary is adapted to the specific tomogram instance, thereby favouring better sparsity and consequently higher quality reconstructions. The reconstruction algorithm is based on an alternating procedure that learns the sparsifying dictionary and employs it to remove artifacts and noise in one step, and then restores the tomogram data in the other step. Simulation and real ET experiments of several morphologies are performed with a variety of setups. Reconstruction results validate its efficiency in both noiseless and noisy cases and show that it yields an improved reconstruction quality with fast convergence. The proposed method enables the recovery of high-fidelity information without the need to worry about what sparsifying transform to select or whether the images used strictly follow the pre-conditions of a certain transform (e.g. strictly piecewise constant for Total Variation minimisation). This can also avoid artifacts that can be introduced by specific sparsifying transforms (e.g. the staircase artifacts the may result when using Total Variation minimisation). Moreover, this thesis shows how reliable elementally sensitive tomography using EELS is possible with the aid of both appropriate use of Dual electron energy loss spectroscopy (DualEELS) and the DLET compressed sensing algorithm to make the best use of the limited data volume and signal to noise inherent in core-loss electron energy loss spectroscopy (EELS) from nanoparticles of an industrially important material. Taken together, the results presented in this thesis demonstrates how high-fidelity ET reconstructions can be achieved using a compressed sensing approach.

Antipodal random sequences with prescribed second-order statistics: application to Compressive Sensing and UWB system based on DS-CDMA

It is usual to hear a strange short sentence: «Random is better than...». Why is randomness a good solution to a certain engineering problem? There are many possible answers, and all of them are related to the considered topic. In this thesis I will discuss about two crucial topics that take advantage by randomizing some waveforms involved in signals manipulations. In particular, advantages are guaranteed by shaping the second order statistic of antipodal sequences involved in an intermediate signal processing stages. The first topic is in the area of analog-to-digital conversion, and it is named Compressive Sensing (CS). CS is a novel paradigm in signal processing that tries to merge signal acquisition and compression at the same time. Consequently it allows to direct acquire a signal in a compressed form. In this thesis, after an ample description of the CS methodology and its related architectures, I will present a new approach that tries to achieve high compression by design the second order statistics of a set of additional waveforms involved in the signal acquisition/compression stage. The second topic addressed in this thesis is in the area of communication system, in particular I focused the attention on ultra-wideband (UWB) systems. An option to produce and decode UWB signals is direct-sequence spreading with multiple access based on code division (DS-CDMA). Focusing on this methodology, I will address the coexistence of a DS-CDMA system with a narrowband interferer. To do so, I minimize the joint effect of both multiple access (MAI) and narrowband (NBI) interference on a simple matched filter receiver. I will show that, when spreading sequence statistical properties are suitably designed, performance improvements are possible with respect to a system exploiting chaos-based sequences minimizing MAI only.

GPU implementation of a hyperspectral coded aperture algorithm for compressive sensing

This paper presents a new parallel implementation of a previously hyperspectral coded aperture (HYCA) algorithm for compressive sensing on graphics processing units (GPUs). HYCA method combines the ideas of spectral unmixing and compressive sensing exploiting the high spatial correlation that can be observed in the data and the generally low number of endmembers needed in order to explain the data. The proposed implementation exploits the GPU architecture at low level, thus taking full advantage of the computational power of GPUs using shared memory and coalesced accesses to memory. The proposed algorithm is evaluated not only in terms of reconstruction error but also in terms of computational performance using two different GPU architectures by NVIDIA: GeForce GTX 590 and GeForce GTX TITAN. Experimental results using real data reveals signficant speedups up with regards to serial implementation.

Parallel hyperspectral coded aperture for compressive sensing on GPUs

The application of compressive sensing (CS) to hyperspectral images is an active area of research over the past few years, both in terms of the hardware and the signal processing algorithms. However, CS algorithms can be computationally very expensive due to the extremely large volumes of data collected by imaging spectrometers, a fact that compromises their use in applications under real-time constraints. This paper proposes four efficient implementations of hyperspectral coded aperture (HYCA) for CS, two of them termed P-HYCA and P-HYCA-FAST and two additional implementations for its constrained version (CHYCA), termed P-CHYCA and P-CHYCA-FAST on commodity graphics processing units (GPUs). HYCA algorithm exploits the high correlation existing among the spectral bands of the hyperspectral data sets and the generally low number of endmembers needed to explain the data, which largely reduces the number of measurements necessary to correctly reconstruct the original data. The proposed P-HYCA and P-CHYCA implementations have been developed using the compute unified device architecture (CUDA) and the cuFFT library. Moreover, this library has been replaced by a fast iterative method in the P-HYCA-FAST and P-CHYCA-FAST implementations that leads to very significant speedup factors in order to achieve real-time requirements. The proposed algorithms are evaluated not only in terms of reconstruction error for different compressions ratios but also in terms of computational performance using two different GPU architectures by NVIDIA: 1) GeForce GTX 590; and 2) GeForce GTX TITAN. Experiments are conducted using both simulated and real data revealing considerable acceleration factors and obtaining good results in the task of compressing remotely sensed hyperspectral data sets.

A fast parallel hyperspectral coded aperture algorithm for compressive sensing using OpenCL

In this paper, we develop a fast implementation of an hyperspectral coded aperture (HYCA) algorithm on different platforms using OpenCL, an open standard for parallel programing on heterogeneous systems, which includes a wide variety of devices, from dense multicore systems from major manufactures such as Intel or ARM to new accelerators such as graphics processing units (GPUs), field programmable gate arrays (FPGAs), the Intel Xeon Phi and other custom devices. Our proposed implementation of HYCA significantly reduces its computational cost. Our experiments have been conducted using simulated data and reveal considerable acceleration factors. This kind of implementations with the same descriptive language on different architectures are very important in order to really calibrate the possibility of using heterogeneous platforms for efficient hyperspectral imaging processing in real remote sensing missions.

l0-deconvolution for compressive diffusion MRI

Diffusion MRI is a well established imaging modality providing a powerful way to non-invasively probe the structure of the white matter. Despite the potential of the technique, the intrinsic long scan times of these sequences have hampered their use in clinical practice. For this reason, a wide variety of methods have been proposed to shorten acquisition times. [...] We here review a recent work where we propose to further exploit the versatility of compressed sensing and convex optimization with the aim to characterize the fiber orientation distribution sparsity more optimally. We re-formulate the spherical deconvolution problem as a constrained l0 minimization.

Recuperação de sinais esparsos. Investigação numérica sobre a quantidade de medidas necessárias para recuperar um sinal esparso

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior

Ultra-low power WSNs: distributed signal processing and dynamic resource management

This thesis presents several data processing and compression techniques capable of addressing the strict requirements of wireless sensor networks. After introducing a general overview of sensor networks, the energy problem is introduced, dividing the different energy reduction approaches according to the different subsystem they try to optimize. To manage the complexity brought by these techniques, a quick overview of the most common middlewares for WSNs is given, describing in detail SPINE2, a framework for data processing in the node environment. The focus is then shifted on the in-network aggregation techniques, used to reduce data sent by the network nodes trying to prolong the network lifetime as long as possible. Among the several techniques, the most promising approach is the Compressive Sensing (CS). To investigate this technique, a practical implementation of the algorithm is compared against a simpler aggregation scheme, deriving a mixed algorithm able to successfully reduce the power consumption. The analysis moves from compression implemented on single nodes to CS for signal ensembles, trying to exploit the correlations among sensors and nodes to improve compression and reconstruction quality. The two main techniques for signal ensembles, Distributed CS (DCS) and Kronecker CS (KCS), are introduced and compared against a common set of data gathered by real deployments. The best trade-off between reconstruction quality and power consumption is then investigated. The usage of CS is also addressed when the signal of interest is sampled at a Sub-Nyquist rate, evaluating the reconstruction performance. Finally the group sparsity CS (GS-CS) is compared to another well-known technique for reconstruction of signals from an highly sub-sampled version. These two frameworks are compared again against a real data-set and an insightful analysis of the trade-off between reconstruction quality and lifetime is given.

Sparse regularization for fiber ODF reconstruction: from the suboptimality of ℓ2 and ℓ1 priors to ℓ0.

Diffusion MRI is a well established imaging modality providing a powerful way to probe the structure of the white matter non-invasively. Despite its potential, the intrinsic long scan times of these sequences have hampered their use in clinical practice. For this reason, a large variety of methods have been recently proposed to shorten the acquisition times. Among them, spherical deconvolution approaches have gained a lot of interest for their ability to reliably recover the intra-voxel fiber configuration with a relatively small number of data samples. To overcome the intrinsic instabilities of deconvolution, these methods use regularization schemes generally based on the assumption that the fiber orientation distribution (FOD) to be recovered in each voxel is sparse. The well known Constrained Spherical Deconvolution (CSD) approach resorts to Tikhonov regularization, based on an ℓ(2)-norm prior, which promotes a weak version of sparsity. Also, in the last few years compressed sensing has been advocated to further accelerate the acquisitions and ℓ(1)-norm minimization is generally employed as a means to promote sparsity in the recovered FODs. In this paper, we provide evidence that the use of an ℓ(1)-norm prior to regularize this class of problems is somewhat inconsistent with the fact that the fiber compartments all sum up to unity. To overcome this ℓ(1) inconsistency while simultaneously exploiting sparsity more optimally than through an ℓ(2) prior, we reformulate the reconstruction problem as a constrained formulation between a data term and a sparsity prior consisting in an explicit bound on the ℓ(0)norm of the FOD, i.e. on the number of fibers. The method has been tested both on synthetic and real data. Experimental results show that the proposed ℓ(0) formulation significantly reduces modeling errors compared to the state-of-the-art ℓ(2) and ℓ(1) regularization approaches.