Browsing by Author "Athaur Rahman bin Najeeb"

An optimal intelligent technique to estimate Time Varying Autoregressive (TVAR) model coefficients is proposed in this thesis. Conventionally, three methods may be used to estimate the TVAR coefficients which are Direct Method (DM), Adaptive Methods (AM) and Basis Function Methods (BFM). All of these methods are built on complex mathematics and recursive in nature which increases the computation time. Although the BFM approach is preferred for few reasons such as (1) they are able to track both slow and fast changing dynamics, (2) BFM does not suffer from convergence problem as in AM. However, it is complex to compute their parameters. In addition to that a type of Basis Function (BF) is able to detect Nonstationary Signals (NSS) with similar characteristics with the BF only. Therefore, limiting the use of AM and BFM in broader NSS processing as naturally they have signal dependent characteristics. In this thesis, a hybrid framework of Artificial Neural Network (ANN) and Genetic Algorithm (GA) known as BP-ANN-GA is proposed to estimate TVAR coefficients. Superior performances of ANN in prediction and its ability to learn complex mapping of input to output is combined with optimization ability of GA to perform this task. Two different ANN architectures are proposed, one to represent TVAR and another one for TVAR BF. These ANN architectures consist of three layers. The number of nodes in input layer is determined by model orders with one hidden layer which consists of an artificial neuron. The third layer has a single node which computes the estimation error which is used to update e the synaptic weight using Backpropagation (BP) learning algorithm. Estimated TVAR coefficients are then fed into GA for further optimization by allowing the TVAR coefficient to be changed within certain limits to ensure the stability. Finally, the TVAR coefficients estimated from proposed method are used to reconstruct various NSS and compared with other methods such as AR, TVAR and BF. It is shown that proposed method yields better accuracy than BP-ANN, AR, TVAR and BF methods. It is also found that the GA optimization produces stable TVAR coefficients when the TVAR coefficients are allowed to be optimized in limits of .Interestingly the BP-ANN-GA also exhibits signal independence characteristics such as independence from model orders and BF, therefore allowing its application to be extended to analyze various types of NSS.

Discrete Fourier Transform (DFT) is a method of choice for reconstructing Magnetic Resonance Images. It is used as basis for reconstructing images in all types of Data Acquisition techniques ranging from conventional techniques such as Projection Method to the latest Parallel Imaging Techniques. Despite the great success of DFT as reconstruction algorithm, it has some inherit limitation. When used with truncated data, it clearly exhibits artifacts. The data set for DFT is expected to be huge, while practically data acquired are small. DFT works best on smooth image function, while practically MRI data set consists of discrete data set. Application of FT on this data leads to artifacts such as truncation artifacts, spikes at discontinuity point of spectrum (Gibbs Ringging). Although spikes could be flatten by applying windows or filters, but this leads to additional problems such as tiny important pathological images could be suppressed as well. To overcome some of these limitations, few alternatives were proposed such as Modified Backprojection Method (MBM), Parametric Methods (PM), Neural Network Methods (NN), Wavelets Methods (WM), etc. Though, clear reconstructed images can be obtained with NN and WM, however, their data representation is doubtful. Studies show during processing several images with motion and a small tumor, NN and WM successfully reconstructed clear images without any indication of motions has taken place. Yet, the pathological tumor has been averaged away as well. Such clinically these methods are invalid and it is a clear indicative of failure. Parametric methods such as Autoregressive (AR) and Autoregressive Moving Average (ARMA) model give positive result. Research and studies conducted during 1980s and 1990s, clearly showing that ARMA can be a good replacement, but due to the computational complexity, it has not been practically implemented and need further clinical testing. Although there are many algorithms available to solve for ARMA, particularly, Transient Error Reconstruction Approach (TERA) were proven be successful. Taking advantage of current microprocessor speeds and successful implementation of previous variety forms of ARMA algorithms, this research tends to implement 2D ARMA on MRI Data set to reconstruct the images. Modified Transient Error Reconstruction Approach (mTERA) method is proposed to solve for 2D ARMA model. Model orders are proposed to be computed dynamically rather than a fixed value as in TERA method. Different approach of AR parameter determination and final image calculation is also been proposed in comparison with TERA method. Raw data for simulation of 2D m-TERA are obtained from Hospital HTAA, Kuantan with courtesy of Philip® Medical System. Images were reconstructed with proposed algorithm using Matlab®. Finally, reconstructed images from 2D m-TERA are compared with standard method 2D IFFT and performance analysis in term of resolution and reconstruction time is presented.