Proton echo planar spectroscopic imaging(PEPSI) is a novel and rapid technique of magnetic resonance spectroscopic imaging(MRSI). To analyze the metabolite in PEPSI by using LCModel, an automatic reconstruction system is necessary. Recently, many researches use graphic processing unit(GPU) to accelerate imaging reconstruction, and Compute Unified Device Architecture(CUDA) is developed by C language, so the programmers can write the program in parallel computing easily. PEPSI data acquisition includes non water suppression and water suppression scans, each scan contains odd and even echoes, these two data are reconstructed separately. The image reconstruction contains k-space filter, time-domain filter, three-dimension fast Fourier transform(FFT), phase correction and combine odd and even data. We use MATLAB, C++ and parallel computing to implement PEPSI reconstruction, and parallel computing applied CUDA which proposed by NVIDIA. In our study, the averaged non water suppression spectroscopic imaging executed by three different programming language are almost the same. In our data scale, the execution time of parallel computing is faster than MATLAB and C++, especially in the FFT step. Therefore, we simulated and compared the performance of one- to three-dimension FFT. Our result shows that accelerating performance of GPU depends on the number of data points according to the performance of FFT and the execution time of single coil PEPSI reconstruction. While the amount of data points is larger than 65536, as demonstrated in our study, parallel computing contribute in terms of computational acceleration.

Deux méthodes d'imagerie par résonance magnétique sont proposées pour analyser in vivo le tissu cérébral de la matière blanche. La première méthode permet l'acquisition ultra-rapide de cartes des métabolites cérébraux par une lecture de l'espace réciproque répétée à des intervalles de quelques millisecondes à l'aide d'une nouvelle trajectoire excentrée, combinée à un gradient de retour. Une procédure de correction de phase, pour prévenir la formation d'artéfacts de repliement dans l'image et le spectre, est introduite sur la base de paramètres déterminés à partir du signal des protons de l'eau. Une acquisition des cartes métaboliques tridimensionnelles de la créatine, de la choline, du N-acétylaspartate, du glutamate et du myo-inositol ont été déterminées de manière fiable dans la substance blanche humaine à 3 Tesla avec une matrice de taille 32 × 32 × 16 et une résolution isotropique de 7 mm. La deuxième méthode permet l'acquisition d'un train de 32 images échantillonnées géométriquement le long d'une courbe de recroissance, en employant une série d'échos de gradient excités par un angle de bascule de 5° pour éviter des effets de saturation. Après transformée inverse de Laplace utilisant une régularisation spatiale, on obtient une distribution continue des temps de relaxation spin-réseau, T1. Dans la région de T1 entre 100 ms et 230 ms, on distingue un pic attribué à l'eau hydratant les membranes de la myéline. La fraction apparente de cette composante de l'eau de myéline augmente en fonction de l'intensité du champ magnétique, de 8,3 % à 3 Tesla, à 11,3 % à 4 Tesla, pour atteindre 15,0 % à 7 Tesla / Two magnetic resonance imaging methods are proposed for the in vivo investigation of human brain white matter tissue. The first method allows the ultra-fast acquisition of maps of brain metabolites by repeating the sampling of k-space at intervals of a few milliseconds, with a center-out trajectory combined with flyback gradients. A phase-correction procedure is introduced to prevent the formation of aliasing artifacts in the image and in the spectrum, on the basis of parameters determined from the signal of the ubiquitous water protons. An acquisition of threedimensional metabolite maps of creatine, choline, N-acetylaspartate, glutamate, and myo-inositol were determined reliably in human brain white matter at 3 Tesla with a 32 × 32 × 16 matrix and a 7-mm isotropic resolution. The second method enables the acquisition of a train of 32 images geometrically sampled along an inversion-recovery curve, using a series of gradient echoes excited by a low 5° flip angle to avoid saturation effects. After inverse Laplace transform, using a spatial regularization, a continuous distribution of the spin-lattice relaxation times, T1, is obtained. In the region of T1 between 100 ms and 230 ms, a small component is attributed to water hydrating myelin membranes. The apparent fraction of this myelin water component increases with the strength of the magnetic field, from 8.3% at 3 Tesla, to 11.3% at 4 Tesla, and 15.0% at 7 Tesla