S.M. Szilagyi

MR brain image segmentation using an enhanced fuzzy C-means algorithm

This paper presents a new algorithm for fuzzy segmentation of MR brain images. Starting from the standard FCM and its bias-corrected version BCFCM algorithm, by splitting up the two major steps of the latter, and by introducing a new factor, the amount of required calculations is considerably reduced. The algorithm provides good-quality segmented brain images a very quick way, which makes it an excellent tool to support virtual brain endoscopy.

Efficient inhomogeneity compensation using fuzzy c-means clustering models

Intensity inhomogeneity or intensity non-uniformity (INU) is an undesired phenomenon that represents the main obstacle for magnetic resonance (MR) image segmentation and registration methods. Various techniques have been proposed to eliminate or compensate the INU, most of which are embedded into classification or clustering algorithms, they generally have difficulties when INU reaches high amplitudes and usually suffer from high computational load. This study reformulates the design of c-means clustering based INU compensation techniques by identifying and separating those globally working computationally costly operations that can be applied to gray intensity levels instead of individual pixels. The theoretical assumptions are demonstrated using the fuzzy c-means algorithm, but the proposed modification is compatible with a various range of c-means clustering based INU compensation and MR image segmentation algorithms. Experiments carried out using synthetic phantoms and real MR images indicate that the proposed approach produces practically the same segmentation accuracy as the conventional formulation, but 20-30 times faster.

Adaptive wavelet-transform-based ECG waveforms detection

A Wavelet-transform-based diverse ECG waveform detection method is presented. An adaptive structure of the processing algorithm can significantly increase the recognition ratio. As a first step, the program will correctly determine the position of QRS complexes and will separate the normal and abnormal beats. Our method allows us to modify in real time the mother-wavelet function, and in this way can be customized to an individual subject or specific waveforms. A parametrical model determines the best performing function for a specific waveform. We used our measurements, but for an adequate comparison with other processing algorithms, tests have been made for the commonly used MIT-BIH database, too. To allow greater waveform diversity we also used our measurements. QRS detection rate was above 99.9%, and for other waveforms the method performs quite well too. The negative influence of various noise types, like 50/60 Hz power line, abrupt baseline shift or drift, and low sampling rate in most cases was almost completely eliminated.

A new method for epileptic waveform recognition using wavelet decomposition and artificial neural networks

The recognition of epileptic waveforms from the electroencephalogram is an important physiological signal processing task, as epilepsy is still one or the most frequent brain disorders. The main goal of this paper is to present a new method to diagnose the epileptic waveforms directly from EEG, by performing a quick signal processing, which makes it possible to apply in on-line monitoring systems. The EEG signal processing is performed in two steps. In the first step, by using the multi-resolution wavelet decomposition, we obtain different spectral components (/spl alpha/, /spl beta/, /spl delta/, /spl theta/) of the measured signal. These components serve as input signals for the artificial neural network (ANN), which accomplishes the recognition of epileptic waves. The recognition rate for all test signals turned out to be over 95%.

On-line QRS complex detection using wavelet filtering

This paper presents a new QRS complex detection algorithm that can be applied in various on-line ECG processing systems. The algorithm is performed in two steps: first a wavelet transform filtering is applied to the signal, then QRS complex localization is performed using a maximum detection and peak classification algorithm. The algorithm has been tested in two phases. First QRS detection in ECG registrations from the MIT-BIH database was performed, which led to an average detection ratio of 99.50%. Then, the algorithm was implemented into a microcontroller-driven portable Holter device. This research is supported by the Hungarian Foundation for Scientific Research, Grant T29830 and FKFP0301/0999 Project.

ECG signal compression using adaptive prediction

A new ECG compression method is presented. First, a prefiltering is effected, followed by a QRS detection. After that the R peaks are localized and the signal is divided into R-R intervals, the original signal can be filtered with less characteristics distortion. This filter is based upon a pondered adaptive long term prediction method. The suggested real-time compression was performed for one of the channels of the MIT/BIH database samples, and can reduce the size of the signal at about 50 bits for one second, without exceeding 10 percent at root mean square reconstruction error (RMSRE). If it is necessary, the algorithm can be used for exact coding, but the size of the concentrated signal highly depends on the sampling rate and resolution. The used adaptive entropy coder introduces about 10 times less redundancy than an optimized Huffman coder.

Heart model based ECG signal processing

Abstract This paper presents the description of the hearts function using a mathematical model, in order to recognize the dangerous states. The developed system applies three different models to obtain the diagnosis: cell model, heart model and chest model. Due to lots of unknown parameters, the computerized “understanding” and simulation of these physiological processes are handled by stochastic processing methods. These methods refer to several physiological problems, like the contraction of the heart, the respiration, the state of the patient, etc. Unfortunately, the printed and electronical literature is often unpunctual, which makes the implementation of such a system harder.

Inference of Evolutionary Jumps in Large Phylogenies using Lévy Processes

Abstract Although it is now widely accepted that the rate of phenotypic evolution may not necessarily be constant across large phylogenies, the frequency and phylogenetic position of periods of rapid evolution remain unclear. In his highly influential view of evolution, G. G. Simpson supposed that such evolutionary jumps occur when organisms transition into so-called new adaptive zones, for instance after dispersal into a new geographic area, after rapid climatic changes, or following the appearance of an evolutionary novelty. Only recently, large, accurate and well calibrated phylogenies have become available that allow testing this hypothesis directly, yet inferring evolutionary jumps remains computationally very challenging. Here, we develop a computationally highly efficient algorithm to accurately infer the rate and strength of evolutionary jumps as well as their phylogenetic location. Following previous work we model evolutionary jumps as a compound process, but introduce a novel approach to sample jump configurations that does not require matrix inversions and thus naturally scales to large trees. We then make use of this development to infer evolutionary jumps in Anolis lizards and Loriinii parrots where we find strong signal for such jumps at the basis of clades that transitioned into new adaptive zones, just as postulated by Simpson’s hypothesis. [evolutionary jump; Lévy process; phenotypic evolution; punctuated equilibrium; quantitative traits.

Biomedical engineering education in Hungary

In medical-biological research, the need for an exact mathematical method is being felt more and more. The Biomedical Engineering Education Programme (BMEEP) in Hungary includes all theoretical processes allowing one to describe the behaviour of a biological system by means of a mathematical model. This paper first deals with the objectives of biomedical engineering training (general principles, etc.); then the BMEEP curriculum is discussed.

Spatial visualization of the heart in case of ectopic beats and fibrillation

This paper presents a dynamic heart model based on a parallelized space-time adaptive mesh refinement algorithm (AMRA). The spatial and temporal simulation method of the anisotropic excitable media has to achieve great performance in distributed processing environment. The accuracy and efficiency of the algorithm was tested for anisotropic and inhomogeneous 3D domains using ten Tusschers and Nygens cardiac cell models. During propagation of depolarization wave, the kinetic, compositional and rotational anisotrophy is included in the tissue, organ and torso model. The generated inverse ECG with conventional and parallelized algorithm has the same quality, but a speedup of factor 200 can be reached using AMRA modeling and single instruction multiple data (SIMD) programming of the video cards. These results suggest that a powerful personal computer will be able to perform a onesecond long simulation of the spatial electrical dynamics of the heart in approximately five minutes.