Loading...
| Friend's email | |
| Your name | |
| Your email | |
| enter code | |
This page was sent successfuly
- Type of Document: M.Sc. Thesis
- Language: Farsi
- Document No: 47388 (05)
- University: Sharif University of Technology
- Department: Electrical Engineering
- Advisor(s): Marvasti, Farrokh; Shamsollahi, Mohammad Bagher
- Abstract:
- Magnetic Resonance Imaging is one of the most advanced medical imaging procedure that noninvasively played in most applications. However, this imaging method is a good resolution, but not in the conventional high speed imaging method, in fact, the main problem is slow. In recent years many studies have been done to accelerate MRI that compressed sensing can be mentioned among them. Such methods, however, have had very good results but MRI systems are very complex. This project investigates the reconstruction of MR images using data from partial non-Cartesian samples aimed at reducing sampling time and also speed up the process of reconstruction of MR images have been studied. In this regard, four algorithms IMAT_DCT, IMAT_WL, ADMM_MRI and ABS_MRI have been introduced. This algorithm is based on the sparseness of DCT and wavelet transform MR images at least one domain in the reconstruction of images of the data in the K partial samples have shown successful performance. IMAT algorithm based on the image in at least one area has become good results. In this context, algorithms IMAT_WL and IMAT_DCT are introduced with the arrival of the DCT and wavelet domains, provides that if the signal is sparse. It also has been shown that these algorithms to IMAT_IMG method assuming that the image within the image is sparse, and accordingly significantly better illustrates the importance of entering into the sphere of its sparse. ADMM_MRI and ABS_MRI algorithms to minimize the energy and software as well as a reconstructed image provided l_1 maintain spatial data K, which have been registered during MR biopsy move. The difference between the two algorithms is how to implement them, so that ADMM_MRI algorithm to solve a constrained optimization problem ADMM by thinning leads us to answer, while ABS_MRI algorithm with repetitive and switching between image area and K to respond us with the least brings energy. It has been shown that minimizing l_1 ADMM_MRI method also leads us to respond with the least energy. The simulation results show that the algorithm ABS_MRI and ADMM_MRI almost of similar performance, but ABS_MRI algorithm is faster and about one-fifth of ADMM_MRI algorithm. The simulation results of the proposed four IMAT_DCT, IMAT_WL, ADMM_MRI and ABS_MRI showed that random sampling of the two Cartesian sampling and groove, the better as well. By applying algorithms on very small samples K to 5% of the samples, showed that the image reconstruction algorithms, we have good quality. The ability to reduce sampling time window opens up very promising because of the need for long-term sampling of the K space resolves. The advantage of the proposed method compared to other methods such as CS, which is less complexity reduces the time and subsequently the cost of shooting more than the other methods. Algorithms are introduced to new methods of reconstruction as DLMRI, DLTG, TLMRI, PBDWS, k _t FOCUSS and Sparse MRI [6,14,43,50,51] during the years 2007 to 2015 have been, not only in terms of quality of image reconstruction are better, but in time also have a much higher speed
- Keywords:
- Iteration Method ; Compressive Sensing ; Reduced Time Magnetic Resonance Imaging ; Partial Non-cartesian Data ; Sparse Image Reconstruction
Digital Object List
-
محتواي کتاب
- view
Bookmark