This paper proposed an automated segmentation algorithm for MR brain images through the complementary use of T1-weighted, T2-weighted, and PD images. The proposed segmentation algorithm is composed of 3 steps. The first step involves the extraction of cerebrum images by placing a cerebrum mask over the three input images. In the second step, outstanding clusters that represent the inner tissues of the cerebrum are chosen from among the 3-dimensional (3D) clusters. The 3D clusters are determined by intersecting densely distributed parts of a 2D histogram in 3D space formed using three optimal scale images. The optimal scale image results from applying scale-space filtering to each 2D histogram and a searching graph structure. As a result, the optimal scale image can accurately describe the shape of the densely distributed pixel parts in the 2D histogram. In the final step, the cerebrum images are segmented by the FCM (Fuzzy c-means) algorithm using the outstanding cluster center value as the initial center value. The ability of the proposed segmentation algorithm to calculate the cluster center value accurately then compensates for the current limitation of the FCM algorithm, which is unduly restricted by the initial center value used. In addition, the proposed algorithm, which includes a multi spectral analysis, can achieve better segmentation results than a single spectral analysis.

We studied the possibility of making an objective diagnosis of dementia based on radiological findings by evaluating cerebral and hippocampal atrophy and the corpus callosum shape on MRI images in patients with Alzheimer-type dementia, compared with healthy elderly individuals. There was a statistically significant difference in the hippocampus area index, the ventricle area index, and the area ratio for the second, forth, and fifth parts of corpus callosum. Discriminant analysis using these three parameters demonstrated the sensitivity of 88.5% and the specificity of 85.7%, suggesting a highly positive diagnostic rate. These results indicate that quantitative MRI measurements could be used for differentiating Alzheimer-type dementia from similar diseases.

In functional brain images obtained by analyzing higher human brain functions using functional magnetic resonance imaging (fMRI), one serious problem is that these images depict false activation areas (artifacts) resulting from image-to-image physiological movements of subject during fMRI data acquisition. In order to truly detect functional activation areas, it is necessary to eliminate the effects of physiological movements of subject (i.e., gross head motion, pulsatile blood and cerebrospinal fluid (CSF) flow) from fMRI time series data. In this paper, we propose a method for eliminating artifacts due to not only rigid-body motion such as gross head motion, but also non-rigid-body motion like the deformation caused by the pulsatile blood and CSF flow. The proposed method estimates subject movements by using gradient methods which can detect subpixel optical flow. Our method estimates the subject movements on a "pixel-by-pixel" basis, and achieves the accurate estimation of both rigid-body and non-rigid-body motion. The artifacts are reduced by correction based on the estimated movements. Therefore, brain activation areas are accurately detected in functional brain images. We demonstrate that our method is valid by applying it to real fMRI data and that it can improve the detection of brain activation areas.

In Vivo Diffusion Tensor Magnetic Resonance Imaging (DT-MRI) can now be used to elucidate and investigate major nerve pathways in the brain. Nerve pathways are constructed by a) calculating a continuous diffusion tensor field from the discrete, noisy, measured DT-MRI data and then b) solving an equation describing the evolution of a fiber tract, in which the local direction vector of the trajectory is identified with the direction of maximum apparent diffusivity. This approach has been validated previously using synthesized, noisy DT-MRI data. Presently, it is possible to reconstruct large white matter structures in the brain, such as the corpus callosum and the pyramidal tracts. Several problems, however, still affect the method's reliability. Its accuracy degrades where the fiber-tract directional distribution is non-uniform, and background noise in diffusion weighted MRIs can cause computed trajectories to jump to different tracts. Nonetheless, this method can provide quantitative information with which to visualize and study connectivity and continuity of neural pathways in the central and peripheral nervous systems in vivo, and holds promise for elucidating architectural features in other fibrous tissues and ordered media.

Wireless mobile communication networks support mobile terminals (MTs) which are free to travel within a service coverage area. In order to locate an MT effectively when a call is initiated, several location management schemes are proposed to keep track of its location. Current methods being used in IS-41 and GSM MAP for location management pose problems such as an increase in signaling traffic in the network, a ping-pong effect and a bottleneck created by signaling traffic at home location register (HLR). In this paper, we propose a new location management scheme that is to reduce the number of location registrations at HLR by making the location registration at HLR unnecessary when an MT moves into neighboring registration area (RA) or when it moves back to the original RA from neighboring RA. Our scheme employs a centralized database architecture using the HLR/VLR concept as in the IS-41 and GSM MAP but alleviates the above-mentioned problems. Analysis results show that our scheme always outperforms current methods and shows an especially better performance at low CMR (call to mobility ratio).

Two isolated pulse models, the Lorentzian-like and the Mixture model, were used to investigate the effect of GMR heads-media with different geometric and magnetic parameters on the readback pulse shape. The matching of these two models with an actual pulse was compared in detail. The dependence of the readback pulse shape of GMR head on the head-media parameters and non-linear distortions was discussed in this paper. When applying these models to evaluate the performance of a recording system, it is necessary to take into account of the difference between the linear superposition of the isolated pulse and the actual readback data pattern. It was suggested to linearize the captured isolated pulse in order to use the model correctly as a useful tool for evaluating the system performance.

Recording mechanism of perpendicular recording was examined using analytical expression of shielded AMR/GMR head response. Pulse shape, roll-off performance, and noise spectra were reasonably explained by the calculated head transfer functions. Comparison with the calculation based on the formulae showed several fundamental characteristics of perpendicular recording: no large media noise at low frequencies in magnetic sense, but simply due to a reflection of a head transfer function: no severe resolution degradation: negligible noise power directly arisen from soft magnetic underlayer. This method will provide a convenient design tool for perpendicular magnetic recording.

Magnetic tunnel junctions (MTJ), i.e., structures consisting of two ferromagnetic layers (FM1 and FM2), separated by a very thin insulator barrier (I), have recently attracted attention for their large tunneling magnetoresistance (TMR) which appears when the magnetization of the ferromagnets of FM1 and FM2 changes their relative orientation from parallel to antiparallel in an applied magnetic field. Using an ultrahigh vacuum magnetron sputtering system, a variety of MTJ structures have been explored. Double Hc magnetic tunnel junction, NiFe/Al2O3/Co and FeCo/Al2O3/Co, were fabricated directly using placement of successive contact mask. The tunnel barrier was prepared by in situ plasma oxidation of thin Al layers sputter deposited. For NiFe/Al2O3/Co junctions, the maximum TMR value reaches 5.0% at room temperature, the switching field can be less than 10 Oe and the relative step width is about 30 Oe. The junction resistance changes from hundreds of ohms to hundreds of kilo-ohms and TMR values decrease monotonously with the increase of applied junction voltage bias. For FeCo/Al2O3/Co junctions, TMR values exceeding 7% were obtained at room temperature. It is surprising that an inverse TMR of 4% was observed in FeCo/Al2O3/Co. The physics governing the spin polarization of tunneling electrons remains unclear. Structures, NiFe/FeMn/NiFe/Al2O3/NiFe, in which one of the FM layers is exchange biased with an antiferromagnetic FeMn layer, were also prepared by patterning using optical lithography techniques. Thus, the junctions exhibit two well-defined magnetic states in which the FM layers are either parallel or antiparallel to one another. TMR values of 16% at room temperature were obtained. The switching field is less than 10 Oe and step width is larger than 30 Oe.

Routing security is related to the confidentiality of the route taken by the data transmitted over the network. If the route is detected by the adversary, the probability is higher that the data are lost or the data can be intercepted by the adversary. Therefore, the route must be protected. To accomplish this, we select an intermediate node secretly and transmit the data using this intermediate node, instead of sending the data to the destination node using the shortest path. Furthermore, if we use a number of secret routes from the starting node to the destination node, data security is much stronger since we can transmit partial data rather than the entire data along a secret route. In this paper, the routing algorithm for multiple secret paths on MRNS (Mixed Radix Number System) Network, which requires O(l) for the time complexity where l is the number of links on a node, is presented employing the HCLS (Hamiltonian Circuit Latin Square) and is analyzed in terms of entropy.

This paper presents a method of MRAC (model reference adaptive control) for MIMO (multi-input multi-output) nonlinear systems using NNs (neural networks). The control input is given by the sum of the output of a model reference adaptive controller and the output of the NN (neural network). The NN is used to compensate the nonlinearity of plant dynamics that is not taken into consideration in the usual MRAC. The role of the NN is to construct a linearized model by minimizing the output error caused by nonlinearities in the control systems.

In this paper, we investigate the bit error rate (BER) of OMC-CDMA (Orthogonal Multi-Carrier-Code Division Multiple Access) with Maximal Ratio Combining (MRC) scheme on two-ray Rayleigh multipath fading channels, and compare the MRC scheme with the Equal Gain Combing (EGC) scheme according to not only the number of users (M) and subcarriers (N) but also the power of delayed signal. We found that the MRC scheme outperforms the EGC scheme, and the BER of OMC-CDMA depends greatly on the power of delayed signal. If the delayed path power is increased from 10% to 50%, as compared with the direct path power, CDMA capacity is decreased approximately by 30% at BER=10-3, N=256 and SNR=15 dB. When the system is heavily loaded by users in large decaying parameter, the performance of OMC-CDMA with the MRC scheme can not increase the BER significantly compared to that with the EGC scheme.

A chemical shift MR method which utilizes a oscillating gradient field is presented in this paper. Frequency modulation resulting from oscillating a gradient field spreads the spectrum that contains both chemical shift and spatial information, over a wide frequency range by using a large modulation factor in FM. The chemical shift spectrum resides within every frequency band segmented by the modulation frequency ωm. The spectral elements gathered from all such frequency segments for a chemical shift frequency contain the spatial image of that particular chemical shift frequency, despite the distortion introduced by a series of the Bessel functions acting as a point spread function. A sum of several Bessel functions of the first kind Jn(. ) is used to approximate the deconvolution process, since the sum staggered with respect to n has a desirable peaking property useful in deconvolution. This leads to devise a new image reconstruction algorithm based on the simple moving average over the spatial coordinate for which the oscillating gradient is applied. Furthermore, the number of echo measurements necessary for an image size of N N is reduced from N2 of the spin echo chemical shift imaging down to N by this method. Simulation results supporting the validity of this method are also presented in this paper.

The origins of baseline shift were discussed considering the measured off-track properties using a wide write head with track widths of 97 µm and a narrow read head with track widths of 2.7 µm. The baseline shift increased when the read head was moved close to the track edge. Beyond the track edge, baseline shift decreased to negative values. The impulse response curve of the MR head to the perpendicular magnetization was estimated from the readback waves of the MIG head and the MR head. The response curve depended on the recorded track width. When the recorded track was narrow, the undershoot of the response curve was smaller than that of the head field based on the 2D double-gap ring head model with infinite track width. This small undershoot induces sensitivity of the DC-component of the recorded magnetization and causes the baseline shift. To calculate the readback waves of the MR head for single-layer perpendicular recording media with narrow-track recording, the effect from stray field at the recorded track edge must be included in the impulse response curve of read head.

This paper proposes a Markov random field (MRF) model-based method for unsupervised segmentation of multispectral images consisting of multiple textures. To model such textured images, a hierarchical MRF is used with two layers, the first layer representing an unobservable region image and the second layer representing multiple textures which cover each region. This method uses the Expectation and Maximization (EM) method for model parameter estimation, where in order to overcome the well-noticed computational problem in the expectation step, we approximate the Baum function using mean-field-based decomposition of a posteriori probability. Given provisionally estimated parameters at each iteration in the EM method, a provisional segmentation is carried out using local a posteriori probability (LAP) of each pixel's region label, which is derived by mean-field-based decomposition of a posteriori probability of the whole region image. Experiments show that the use of LAPs is essential to perform a good image segmentation.

Linear signal analysis (LSA) is the conventional method of estimating the playback voltage and pulse width in linearly operating shielded GMR heads. To improve the accuracy of LSA, a new, highly precise LSA which includes the effect of the magnetization distribution in the medium and inhomogeneous biasing by domain control magnets, was developed. Utilizing this new LSA to calculate the playback waveforms, the calculated peak voltage and pulse width were compared with the experimental values and agreement within 10% was obtained. As the result of estimation using the new LSA, it is considered that the use of a vertical-type spin-valve head will make it possible to achieve a recording areal density of 40 Gb/in2.

A method for locating the minimum eigenvalue and its corresponding eigenvector is considered. The core procedure utilized is the modified Rayleigh quotient iteration (MRQI). The convergence rate of the Rayleigh quotient iteration (RQI) is cubic. However, unfortunately, the RQI may not always locate the minimum eigenvalue. In this paper, a new MRQI that can always locate the minimum eigenpair is given. Based on the MRQI, a fast algorithm to locate minimum eigenpair will be proposed. This method has the following characteristics. First, it does not compute the inclusion interval. Second, it works for any Hermitian matrix as well as Toeplitz matrix. Third, it works on matrices having more than one minimum eigenvalue. Fourth, the numerical error of this method is very small. Fifth, it is attractively simple and fast. The convergence rate of this method is asymptotically cubic. MATLAB simulation results show that this method may outperform other methods. The term MRQI has been already used. Differences in several MRQI methods are discussed. Mathematical properties of the MRQI are investigated. This research can be effectively applied to diverse field of the signal processing including communication, because the signal space can be efficiently obtained.

This paper presents a feature extraction model MAGNET' to find the deepest point of branched sulcus. Our model demonstrates magnetic principle and consists of four types of ideal magnetic poles: an N-pole and three S-poles. According to attractive or repulsive Coulomb forces between their poles, one of the S-poles is pushed to the deepest point of the sulcus. First, we explain our model on the simple sulcus model. Second, we apply it to the sulcus with implicit branches. Our model can detect the target points in every branch. Then an example to realize the model on a synthetic image is introduced. We apply our model to human brain MR images and human foot CT images. Experimental results on human brain MR images show that our method enable us to successfully detect the points.

One serious problem affecting the rest and active state images obtained during a functional MRI (fMRI) study is that of involuntary subject movements inside the magnet while the imaging protocol is being carried out. The small signal intensity rise and small activation areas observed in the fMRI results, such as the statistical maps indicating the significance of the observed signal intensity difference between the rest and active states for each pixel, are greatly affected even by head displacements of less than one pixel. Near perfect alignment in the subpixel level of each image with respect to a reference, then, is necessary if the results are to be considered meaningful, specially in a clinical setting. In this paper we report the brain displacements that take place during a fMRI study with an image alignment method based on a refined crosscorrelation function which obtains fast (non-iterative) and precise values for the inplane rotation and X and Y translation correction factors. The performance of the method was tested with phantom experiments and fMRI studies using normal subjects executing a finger-tapping motor task. In all cases, subpixel translations and rotations were detected. The rest and active phases of the time course plots obtained from pixels in the primary motor area were well differentiated after only one pass of the motion correction program, giving enhanced activation zones. Other related areas such as the supplementary motor area became visible only after correction, and the number of pixels showing false activation was reduced.

Segmenting the images obtained from magnetic resonance imaging (MRI) is an important process for visualization of the human soft tissues. For the application of MR, we often have to introduce a reasonable segmentation technique. Neural networks may provide us with superior solutions for the pattern classification of medical images than the conventional methods. For image segmentation with the aid of neural networks of a reasonable size, it is important to select the most effective combination of secondary indices to be used for the classification. In this paper, we introduce a vector quantized class entropy (VQCCE) criterion to evaluate which indices are effective for pattern classification, without testing on the actual classifiers. We have exploited a newly developed neural tree classifier for accomplishing the segmentation task. This network effectively partitions the feature space into subregions and each final subregion is assigned a class label according to the data routed to it. As the tree grows on, the number of training data for each node decreases, which results in less weight update epochs and decreases the time consumption. The partitioning of the feature space at each node is done by a simple neural network; the appropriateness of which is measured by newly proposed estimation criterion, i. e. the measure for assessment of neuron (MAN). It facilitates the obtaining of a neuron with maximum correlation between a unit's value and the residual error at a given output. The application of this criterion guarantees adopting the best-fit neuron to split the feature space. The proposed neural classifier has achieved 95% correct classification rate on average for the white/gray matter segmentation problem. The performance of the proposed method is compared to that of a multilayered perceptron (MLP), the latter being widely exploited network in the field of image processing and pattern recognition. The experiments show the superiority of the introduced method in terms of less iterations and weight up dates necessary to train the neural network, i. e. lower computational complexity; as well as higher correct classification rate.

A modified fully-digital code tracking loop is proposed in this paper for direct-sequence spread-spectrum signaling on a frequency-selective fading channel. A data-modulated channel estimator is used to cope with the time-varying Rayleigh fading effect and the data modulation effect, and extract the desired error signal from each path independently in the multipath environments. By taking advantage of the inherent diversity with the maximal ratio combining (MRC) or a proposed Even/odd maximal ratio combining (EMRC) technique, this modified code tracking loop can avoid the problem due to the drift or flutter effects of the error characteristics, and provide better performance on frequency selective fading channels. Extensive computer simulation has verified the analysis and indicated very attractive behavior of the proposed digital tracking loop.