Discrete and Continuous spatial encoding methods are described with details of understanding principles and practical implications. Step by step experimental op- timization procedure of these methods to achieve slice selection are also discussed. In the subsequent chapters we use these methods for different applications. Spin-lattice relaxation parameters of NMR active nuclei provide valuable infor- mation on molecular dynamics. Single scan selective excitation methods of mea- surement of T1 result in significant reduction of time compared to the standard inversion recovery method and are attractive tools of applications in `Real time' NMR investigations of biological and chemical processes. It is shown here that the addition of the gradient echo following the selective excitation not only significantly improves the S/N ratio, but also makes GESSIR a versatile pulse sequence. Using this sequence, T1 values ranging from 2 s to 56 s have been measured with accuracy comparable to the standard IR experiment. This indicates that it is possible to utilize GESSIR for a wide range of molecules containing protons and hetero nuclei with medium to long T1 relaxation times as a routine NMR technique. The utility of the technique for studying other relaxation parameters has also been demonstrated. It may be mentioned that for measurement of relaxation parameters routinely, a few well-chosen points are enough. A fine selection of large number of experimental points could be useful when high accuracy is required or Chapter 3. GESSIR 91 for applications where certain property of the system investigated is changing in a continuous manner spatially and requires large number of slices to be selected as discussed in the next chapter. The long duration of time-honored two dimensional experiments is reduced to fraction of seconds by employing the ultrafast encoding experiments. Main com- plications in making the UF experiments available for routine use were the limited spectral widths and resolution in both UF and conventional dimensions. Various developments have been made in the path of improvements in increasing the spectral width in UF dimension. Of these, two experimental methods that are already proposed, namely the folding of peaks into the observable spectral window and the interleaved acquisition which doubles the spectral widths in both dimensions. The integration of covariance processing with ultrafast technique yields better correlated spectrum with considerable improvement in resolution. The effectiveness of the new processing is demonstrated for UF COSY experiments which can be easily extended to other ultrafast homonuclear experiments like TOCSY, NOESY as well as multi dimensions. The proposed processing method is an initial step to work on improving resolutions of UF data and making the ease of applicability of ultrafast spectroscopy as a routine multidimensional NMR. At the same time of this work W. Qui et.al [268] proposed a processing method based on covariance and pattern recognition for improving resolutions of spatially encoded data. They used pattern recognition algorithm also for avoiding the artifacts due to very low resolution data available with the UF experiment. They implemented the method UF TOCSY spectra and shown resolution improvement with the covariance pro- cessing for relatively more number of detection gradients which is many times hardware limited. Our method of covariance data processing is essentially same as that of Qui, less number of acquisition gradients were used in our processing, linear prediction and apodization concepts were utilized and the artifacts arise due mismatch of datas with positive and negative acquisition gradients are minimized by shifting one the data. In conclusion new methods of processing/the combination of various processing methods of the ultrafast data have the scope of improving the quality of ultrafast NMR spectra.

La spectroscopie de résonance magnétique nucléaire (RMN) est un outil puissant qui permet l’étude directe de mélanges de manière non destructive. Les spectres RMN de petites molécules en solution peuvent être différenciés grâce à la stratégie DOSY (diffusion-ordered spectroscopy), une méthode de ‘chromatographie virtuelle’ qui s’appuie sur la mesure de coefficients de diffusion translationnelle. Les principaux obstacles à l’utilisation de la DOSY sont liés à la piètre sensibilité de la RMN de manière générale mais aussi à la nécessité d’introduire une dimension temporelle supplémentaire d’acquisition, ce qui va augmenter de manière significative la durée de l’expérience. Ce travail de thèse a pour objectif de mettre au point des outils inédits de RMN plus rapides et plus adaptés à la caractérisation de mélanges peu concentrés de petites molécules. Dans un premier temps, le concept de codage spatial de la diffusion dans l’expérience DOSY a été généralisé. Mis à profit dans les méthodes RMN dites ‘ultrarapides’, l’utilisation d’une dimension spatiale plutôt que temporelle pour encoder le phénomène de diffusion permet une accélération des expériences de RMN multidimensionnelles de plusieurs ordres de grandeur. L’acquisition séquentielle de spectres est remplacée par une acquisition parallèle de ces spectres dans différentes parties de l’échantillon. L’étude poussée des méthodes de DOSY rapides s’est appuyée sur des outils de simulation numérique dans le but d’améliorer la résolution des spectres et la précision des résultats. Les problèmes de sensibilité ont été abordés via le couplage des méthodes DOSY rapides avec des méthodes d’hyperpolarisation qui permettent d’augmenter l’intensité du signal. La combinaison des méthodes de diffusion conventionnelles avec les méthodes avancées de RMN ultrarapide et d’hyperpolarisation permettront des avancées significatives pour l’analyse de mélanges, en particulier les mélanges dynamiques. / NMR spectroscopy is a powerful tool that allows a direct study of mixtures in a non-invasive manner. The NMR spectra of molecular species in mixtures can be separated with diffusion-ordered spectroscopy (DOSY), a ‘virtual chromatography’ approach based on the measurement of translational diffusion coefficients. Major limitation of DOSY comes from the time-dependent diffusion dimension, which results in long experiment durations, and also from the low sensitivity of NMR. The present work aims to build an innovative tool for mixtures characterization that will be faster and more efficient for low concentrated samples. We first generalized the concept of nD spatially encoded (SPEN) DOSY experiments for the analysis of complex mixtures. As bring forward by the so-called “ultrafast NMR” (UF NMR), the use of a spatial dimension to encode diffusion can accelerate experiments by several orders of magnitude since it replaces the sequential acquisition of sub-experiments by a parallel acquisition in different slices of the sample. More advanced exploration of SPENDOSY were carried out using numerical simulations for purpose of resolution and accuracy improvement. To address sensitivity issues, we then demonstrated that SPENDOSY data can be collected for hyperpolarized substrates. This particular coupling between conventional diffusion-based method with advanced techniques such as ultrafast NMR and hyperpolarization should mark a significant progress for complex mixtures analysis especially for time-evolving processes.

La Spectroscopie de Résonance Magnétique (SRM) constitue un outil non-invasifunique pour l’exploration biochimique du métabolisme des organismes vivants. Cependant,en raison des champs magnétiques couramment utilisés chez l’homme etle petit animal, la SRM in vivo du proton ne permet pas de quantifier précisémentla concentration de tous les métabolites présents dans le cerveau. La SRM à deuxdimensions spectrales (SRM 2D), technique utilisée en routine en chimie, permetde séparer efficacement les signatures spectrales des métabolites facilitant ainsi leuridentification et leur quantification en termes de concentrations. Les travaux réalisésdans le cadre de cette thèse concernent le développement de méthodes d’acquisitionet de quantification de spectres RMN 2D J-résolus in vivo et sont présentéssuivant deux axes majeurs. Le premier axe concerne les travaux relatifs à la SRM2D J-résolue conventionnelle qui ont fait l’objet du développement d’une séquenceJ-PRESS sur un imageur 7 T pour l’acquisition de spectres 2D sur le cerveau de rat.Les données acquises sont traitées avec une méthode d’analyse spectrale développéeet optimisée spécifiquement pour la quantification de données SRM 2D J-résolues,reposant sur une connaissance a priori et un ajustement numérique dans le domainetemporel. Le second axe concerne les travaux relatifs à la réduction de la duréed’acquisition en SRM 2D avec le développement de techniques basées sur le conceptrécent de RMN ultrarapide. Une nouvelle séquence de SRM 2D J-résolue ultrarapidea été développée et validée sur un imageur 7 T et a permis l’acquisition de spectres2D complets avec une durée d’acquisition de l’ordre de la seconde. / In vivo proton Magnetic Resonance Spectroscopy (MRS) is a powerful tool for metabolicprofiling because this technique is non-invasive and quantitative. However,conventional localized spectroscopy presents important in vivo metabolic informationthrough overlapped spectral signatures greatly affecting the quantification accuracy.Two-dimensional (2D) MRS, originally developed for analytical chemistry,has great potential to unambiguously distinguish metabolites. Therefore, metabolitequantification is improved allowing accurate estimation of their concentrations. Inthis thesis, the research findings are presented under two main headings. The firstline of research focuses on conventional 2D MRS J-resolved. A J-PRESS sequencewas developed allowing the acquisition of in vivo 2D MRS spectra, which were processedby a dedicated quantification method. Experiments were performed on therat brain using a 7 T imaging system and different sampling strategies were evaluated.The quantification method, specifically developed to handle 2D J-resolved MRSdata quantification in time domain, is based on a strong prior-knowledge. However,2D MRS suffers from long acquisition times due to the collection of numerous incrementsin the indirect dimension. Therefore, the second line of research focuseson the reduction of acquisition time using recently developed methods based on theultrafast NMR concept. A new pulse sequence was designed, allowing 3D localizedultrafast 2D J-resolved spectroscopic acquisition on a 7T small animal imaging system. This breakthrough allows the acquisition of a complete 2D spectrum in a singlescan, resulting in acquisition times of a few seconds.