Information Technologies in Biomedicine

The paper discusses the alternative method of medical experts participation in technical inventions for medicine. Blind tests and various statistic-based correlations of human and automatic interpretation results are commonly used today. Our paper postulates a deeper insight into the expert performance in order to better understanding and simulating his reasoning in the software. The benefit is twofold: the measurement is objective and the closer simulation of human reasoning yields better performance in case of unexpected input. Although the area of application is the very broad intersection of medicine and technology, we focus on the automatic ECG interpretation, and propose the agile software featuring a human-like behavior. Two examples of experiments aimed at extraction of some aspects of ECG interpretation knowledge are also included in the presentation.

The paper discusses application issues of using the metal implants for treatment of the cardiovascular diseases. The analysis of the biophysical conditions of the heart-coronary vessels system has been used to distinguish the tissue environment properties which should be compatible with properties of the metal biomaterial and stent surface. The need to determine the correct quality and service properties of the coronary stents has been indicated, which refer first of all to their design form, physical and chemical properties of the metal biomaterial and its surface. Based on that the Author of the work has proposed his own methodology for forming and controlling the service properties of the stents. It takes into account the required relationships between structure, and mechanical properties of the stent biomaterial, and the physical and chemical properties of its surface - adjusted to the specific features of the cardiovascular system.

The role of computers in orthopedic research and education and clinic is expanding rapidly. The computer assisted methods and modern technologies lately are changing musculoskeletal diagnostics, orthopedic surgery and rehabilitation. Technologies for computer assisted surgery (CAS) at the beginning were introduced into surgical practice for pre-operative planning and to enhance accuracy and safety for a variety of procedures. The introduction of highly demanding complex surgical procedures requires better visualization and detailed anatomy recognition intraoperatively. The new abilities to manipulate images during pre-operative planning increase an accuracy of surgical procedures. The orthopedic surgeon needs to be aware that technology driven methods are feasible and suitable nowadays. Computer-assisted methods for orthopedic surgery utilize the use of computers and robotic technology to assist in providing musculoskeletal care. Since its clinical implementation in neurosurgery, computer assisted methods in orthopedic surgery namely: surgical navigation, CAOS, CAD, distant learning, rapidly evolve in numerous applications. The final integration of all computerized applications creates new level of orthopedic surgical workflow of digital data that can be named Orthopedic PACS. Mentioned above methods have some clinically relevant implementations already, but further development is expected. The orthopedic surgeon should be aware of advantages as well pitfalls of its use. Clear understanding the goals, applications, and limitations of the computerized methods determine its successful current and future clinical use for the further improvement of the patients care. The future systems for daily practice should be characterized by easy learning, intuitive and friendly in use, and foolproof. The orthopaedic surgeon who understands and applies computerized technologies can expect further improvement in patients care.

This paper presents a status of the computer-aided diagnosis (CAD) in a context of nowadays challenges and limitations of advanced digital technologies in radiology, medical imaging systems and networked medical care. Computer-assisted interpretation of radiological examinations requires flexible more/less formal image modeling, reliable numerical descriptors of diagnostic content, indicators of image accuracy, and first of all effective methods of image understanding. Moreover, it is important to base the computer assistance design on understanding of human determinants of diagnosis, characteristics and enhancements of observer performance and dynamic platform of medical knowledge – a formal description of semantic image content, i.e. ontology. To make it fully useful, computer-based aid tools are integrated into networked radiology environment based on PACS/RIS/HIS/teleradiology systems interfaced to diagnostic workstations. The general concepts of CADs were exemplified inter alia with breast cancer and brain stroke diagnosis applications.

It is considered the problem of representation of irregular structures, typical in biomedical images, by morphological spectra. Some basic properties of morphological spectra have been reminded. Then several possible methods of morphological spectra presentation are illustrated by a numerical example. The problem of representation of selected classes of irregular structures is considered and illustrated by the examples of vertically elongated, compact and branching structures representation. It is shown that for each class of irregular structures a hierarchy of spectral components indicating their role in representation of the class can be statistically established.

Potential contours are methods for automatic image analysis. They can be interpreted as contextual classifiers that use expert knowledge and operate in supervised or unsupervised mode. In the present paper, potential contours adapted in the supervised way are examined on medical images.

Potential active contours are based on the well known potential function method of classification, where the label assigned to the object depends on the distribution of other known and already classified objects. In the classic formulation of this approach, the known objects are fixed, while in the potential active contours’ method their position and parameters are subject of optimization. In the present paper, the basic concept of potential active contours is described and its features are presented.

Image magnification takes important part during medical examination. It is especially very helpful in observing small details and their manual measurements. There are a number of magnification methods, however most of them cause image distortions. The paper presents a fractal magnification method, which allows minimizing these negative effects besides the blocks effect.

In the current study a modification of the live-wire approach to image segmentation has been developed. A Fuzzy C-Means (FCM) clustering procedure has been implemented before the cost map function is defined. This shrinks the area to be searched resulting in a significant reduction of the computational complexity. The method has been employed to Computed Tomography (CT) studies. Segmentation of the abdomen structures has been performed in order to evaluate the method.

In the article a new method of automatic image segmentation is presented. The aim was to eliminate the necessity of defining the number of outcome areas. Homogeneous areas take part in the growth process. The areas merge when the homogeneousness condition is fulfilled. The threshold value changes during the segmentation process, fitting the changeable conditions.

This paper concerns content-based image retrieval in medical domain considering the challenges of rapidly growing amounts of medical data, permanent progress of computer-aided radiology and the development of global data exchange networks (like Mammogrid). The aim of featured research was to propose effective content-based image retrieval algorithms in two approaches to that problem. The first is medical images indexing for various modalities, on the assumption that we do not attempt to analyze image semantics. In that approach we try to find the images ’visually similar’ to the given one – with similar organ, modality, orientation, etc. The second approach undertaken in conducted research is medical images indexing with taking into consideration their semantics. Such approach makes use of ’domain knowledge’ about specified modality, examination, giving the opportunity to introduce descriptors of image semantics, especially related to diagnostic content. The methods and algorithms characterized in this paper are both related to various modalities and strictly dedicated to the one modality only - in this research it is mammography. The obtained results show the usefulness of proposed methods.

The huge amount of digital images generated in hospitals leads to the need of automatic storage and retrieval of them. A

Picture Archiving and Communication System

(PACS) should incorporate properties allowing to retrieve these images and adding

Content-Based Image Retrieval

(CBIR) capabilities to PACS makes it more powerful to assist diagnosis. Such systems provide features which combine color, shape and spatial features to query an image. In response to a user’s query, the system returns images that are similar in some user-defined sense. Our purpose in this study is to develop a method of image mammogram feature extraction (microcalcifications and masses features) in CBIR system.

One of the important attributes of cancerous masses is their malignancy as it suggests a rapid growth of the cancer and possibility of metastasis. Malignancy, which denotes a special pathology of the tissue, is closely related to the existence of quasi-linear structures (spicules) emanating from the central mass. Hence, the tasks of malignancy and spicularity assessment are very often treated jointly. We propose a novel set of features enriching already existing pool of features for classification of masses. Our features are based on simple MM operations, pixel counting, and some basic algebra. To be more specific, given a contour of a cancerous mass we compute a sequence of dilations, and then count the number of pixels on the inner and the outer contour of each dilation. The contour pixel numbers are plotted against the size of the disk-shaped structuring element. The MM features are calculated from the plot via simple algebraic operations. The crucial point is that all the features are zero iff the input contour is circular. This distinctive property forms a basis for successful classification with the

A

z

values higher than for the features existing in the literature. The additional advantage of our approach is the simplicity of the proposed features. In contrast to many features found in the literature, no sophisticated algorithms are employed, so reimplementation of the features should be easy for anyone interested.

The computer-assisted support of acute ischemic stoke detection was the subject of our research reported in this paper. The conditioning of early stroke diagnosis based on CT examinations was analyzed. The multiscale extraction of the subtlest signs of hypodensity which were often undetected in standard CT scan review was presented. Proposed method was as follows: evidence-based description of ischemic changes, the analysis of hypodensity signs across scales, noise reduction and hypodensity extraction, and following display of ischemic changes localized in source brain image space. Important issues were: –extension of the brain tissues for marginal and missing space after deskulling and segmenting of unusual areas; –multiscale transform selection; –denoising in scale-space domain; –visualization conditions fixing. Three forms of extracted stroke sings visualization were proposed. Increased visibility of cerebral ischemia for difficult-in-diagnosis cases was experimentally noticed.

In the current study, a fuzzy-connectedness-based approach to fine segmentation of demyelination lesions in Multiple Sclerosis is introduced as an enhancement to the existing ‘fast’ segmentation method. First a fuzzy connectedness relation is introduced, next a short overview of the ‘fast’ segmentation method is presented. Finally, a novel, automated segmentation approach is described. The combined method is applied to segmentation of clinical Magnetic Resonance FLAIR Images.

The subject of the paper is the evaluation of a brain atrophy depending on computed tomography (CT) images. There are two computer aided methods under investigation. The first one is a semi-automatic volumetric method, the second is a fully automatic one and depends on the fractal dimension. Results of the experiment with 68 patients show that both methods are able to estimate the brain atrophy very similarly to the evaluation made by expert radiologist with the coefficient of correlation in the range of 0.7-0.9. The great advantage of the automatic method comparing to volumetric method is generating results in a very short time, that counts in ordinary clinical practice and could be used to experiment with huge patient groups.

This paper shows an automatic registration method of the A- and B-group of MRI brain for the same patient. The time difference between recording the A-group and the B-group is equal six months. The automatic registration of the A- and B-group of MRI brain is based on the entropy and energy measures of fuzziness. First, two sequences (A- and B-group) are converted to a fuzzy representation. Then, the entropy and energy measures are employed in the NCC and GD methods. The alignment based on energy and entropy fuzzy measures shows a significant improvement in comparison with the implementation of the original image.

Fractal analysis is a reasonable choice in applications where natural objects are dealt with. Fractal dimension is an essential measure of fractal properties. Differential box-counting method was used for fractal dimension estimation of radiological brain images. It has been documented in this paper that this measure can be used for automatic classification of normal and pathological cases.

Atheromatous narrowing and subsequent occlusion of the coronary vessel cause coronary artery disease. Application of optimised feed forward multi-layer back propagation neural network (MLBP) for detection of narrowing in coronary artery vessels is presented in this paper. The research was performed using 580 data records from traditional ECG exercise test confirmed by coronary arteriography results. Each record of training database included description of the state of a patient providing input data for the neural network. Level and slope of ST segment of a 12 lead ECG signal recorded at rest and after effort (48 floating point values) was the main component of input data for neural network. Coronary arteriography results (verified the existence or absence of more than 50% stenosis of the particular coronary vessels) were used as a correct neural network training output pattern. More than 96% of cases were correctly recognised by thoroughly verified MLBP neural network. Leave one out method was used for neural network verification so 580 data records could be used for training as well as for verification of neural network.

In this paper we present first results of our, computer aided, research into angiogenesis process, conducted in association with radiologists from local clinical hospital. Presented here is its informatics part, which was to estimate, basing on CT scans, the quotient of the tumour volume to the number of its capillary veins. Should some correlation be found between the effectiveness of the cancer healing process and the quotient, it would mean that healing is affecting the angiogenesis process.

Angiogenesis is a process of forming new blood vessels from the already existing ones, and is the main cause of violent cancer development. Let us say that cancer cells force the angiogenesis process, thus making vasculature nourishing cancer cells a colony and enabling its growth. Without it, a tumour would be harmless. It is no wonder that modern medicine is trying hard to develop a method to stop the angiogenesis process. If during treatment the number of capillaries decreases, the tumour is less effectively nourished and the disease recedes, furthermore the ability of the cancer to spread over the body is being limited.

The crucial part of the lung cancer computer-aided diagnosis (CAD) is the segmentation of pulmonary nodules in Computed Tomography (CT) study. A new multilevel approach based on fuzzy connectedness principles has been developed. The three-dimensional fuzzy connectedness analysis requires a dedicated preprocessing stage in order to limit the computation time to a reasonable range. It consists of the initial thresholding, connected components labeling, and creating the binary masks of regions within the thorax. For nodules connected to pleura or vessels, a separation step is needed, using mathematical morphology and the shape analysis. Separation of the nodule and pleura is performed in the preprocessing stage, whereas separation of a nodule and connected vessels – in the postprocessing stage. In this paper the methodology is described and illustrated.

The whole segmentation method has been tested on a set of three-dimensional CT images of the thorax with delineated lung nodules. Results and some examples of such an application are shown.

In the present study the goal was to quantify the stresses acting locally on pulmonary epithelial cells in order to better understand the dynamic behavior of these cells. To quantify mechanotransduction responses, one must first understand the magnitude and distribution of stresses on the epithelial cells. It was investigated a two-dimensional, mathematical model of airway reopening, using a flow-driven semi-infinite bubble progressing through an airway as it clears a liquid occlusion was created. The flow in this system was highly viscous, and thus was governed by Stokes equations. This 2-D model was solved computationally using the boundary element method (BEM) in conjunction with lubrication approximations. Algebraic expressions were developed that could be used simply and accurately predict the fluid stress based upon the fluid viscosity,

μ

, channel height,

H

, cell size,

A

, and flow-rate,

Q

. From the solution, it was determined the stationary-state stresses acting on the epithelial cells. The results indicated that the magnitude of both the

x

- and

y

-stresses acting on the walls’ cells were directly related to the cell protuberance topography and produced a complex stress field.

This paper aims at quality assessment and control of bronchoscopy video data used for diagnosis and managed in medical information systems. Application of different video coding/decoding methods according to MPEG and JPEG family standards were considered for effective video data storage and communication. Enhanced image sequences preview, analysis and interpretation according to radiological procedures is possible because of improved, flexible, hierarchically ordered data representation. Useful coding or transcoding algorithms were studied, experimentally optimized and verified according to quality preserving criteria based on objective numerical measures, subjectively controlled. Compression schemes suitable for bronchoscopy video were concluded.

In this paper the fuzzy rule-based system for nuclei classification is presented. Firstly, in order to receive proper partition of objects (nuclei) the definition of features which can be used for diagnosis of laryngeal pathology based on contact endoscopy images is described. After the feature selection the fuzzy clustering process is realized. It creates the set of training input-output data pairs which later are used for generation of fuzzy rules by means of the method called learning from examples.

In this paper, the development of a new module of Internet Information System for synthesis of melanocytic skin lesion images is briefly outlined. The key approach in the developed synthesis methodology of images is a semantic conversion of textual description of melanocytic skin lesions - by an in-house developed system - into digital images. It was found, that the developed methodology can be successfully used in the process of teaching of dermatology students and also in training of preferred medical doctors. The system is available via Internet at the website http://www.melanoma.pl.

In the paper we present an algorithm for the identification of retina layers using the Canny edge detection for images obtained with OCT (Optical Coherence Tomography) Copernicus. The developed algorithm is an extension of the approaches covered in [4] and allows the identification and detection of hyaline-retinal border layers, the retina and other. The algorithm was implemented in the Matlab environment and the C language.

The paper discusses various aspects of diagnostic quality estimate in an adaptive ECG interpretation system. Since the ECG interpreting software adapts to the patient status and diagnostics goal variations, the commonly used term of data quality must be revised. We propose backgrounds and demonstrate the use of a multidimensional quality hyperspace depending on the assumed system adaptability. The emerging need of appropriate reference database and standardized description of human experts behavior is also addressed here. The subset of proposed quality estimators is used as adaptivity criteria in a prototype of wireless monitoring system for cardiology. The paper concludes with guidelines for testing the processing performance and with a comparison of rigid and adaptive software features.

Automatic measurement of repolarization duration is highly prone to errors. The paper is focused on the problem of ECG noise suppression prior to the measurement. Two methods are compared: the traditional linear lowpass filtering and the projective filtering of the time aligned ECG beats. The performed experiments show that the projective filtering has much stronger impact on the precision of the measurements.

Noises which appear during recording of biomedical signals are seldom characterized by the ideal Gaussian model, because noises can have an impulsive nature. In this work the robust filter is introduced on the base of the Gaussian radial basis function. For that purpose the distance measure in the feature space is defined. Minimizing of this distance measure allows to estimate output value in a moving window filter. Presented filter was applied to suppress muscle noise in ECG signal. Some properties of presented filter are also shown. The experiments performed show that introduced filter is effective in suppression of noise of biomedical signals.

Averaging signals in time domain is one of the main methods of noise attenuation in biomedical signal processing. This paper proposes a new weighted averaging method using criterion function minimization and based on partitioning of input set in time domain, which is a generalization of an existing method. Performance of the new method is experimentally compared with the traditional averaging by using arithmetic mean and another weighted averaging method based on criterion function minimization.

The analysis of the electrocardiographic signal recordings is greatly useful in the screening and diagnosing of cardiovascular diseases. However usually recording of the electrical activity of the heart is performed in the presence of noise. One of the commonly used techniques to extract a useful signal distorted by a noise is weighted averaging, since the nature of ECG signal is quasi-cyclic with level of noise power varying from cycle to cycle. This paper proposes a new weighted averaging method, which incorporates empirical Bayesian inference and the expectation-maximization technique. It is an extension of an existing method by introducing Cauchy distribution and the unknown parameter is estimated using interquartile range. Performance of the new method is experimentally compared with the traditional averaging by using arithmetic mean and other empirical Bayesian weighted averaging methods.

In this paper we presented an approach to segmentation of an electrooculography (EOG) signal. For segmentation we have used the elements of the fuzzy set theory. Results obtained in our numerical experiments show usefulness of proposed approach. Our method can be also used for the generating of a learning set necessary for the neural networks or the fuzzy–neural systems training.

This paper presents some aspects of a on-line remote fetal monitoring system based on the GPRS packet data transmission service and WAN network. The system enables analysis, archiving and presentation of signals directly from homes of high-risk pregnancy patients. Since the transmitted signals have to be presented on-line in the surveillance center, possible problems in transmission channel were considered in details. Data stream continuity for visualization is ensured by implementation of incoming data buffer and procedures of its data read-period control.

Cardiotocographic monitoring (CTG) is the primary biophysical method for assessment of the fetal state. It consists in analysis of fetal heart rate variability, uterine contraction activity and fetal movements signal. Visual analysis of printed cardiotocographic traces is difficult so the computerized fetal monitoring systems are a standard in clinical centres. In the proposed work we investigated the ability of the application of artificial neural networks for the prediction of newborn sex using parameters of quantitative description of CTG traces. We examined the influence of input data representation (numerical or categorical) and the influence of the gestational age on the classification quality. We obtained the classification quality at a level of 80% and therefore we may state, that there is rather a strong relation between the fetal gender and the fetal heart rate variability.

Estimation of the instantaneous variability of the fetal heart rate (FHR) is affected by the autocorrelation techniques commonly used in Doppler ultrasound channel of today’s fetal monitors. Considerably decrease of short-term variability have been noted, which is quite surprising because the fetal monitors determine the fetal heart rate with quite satisfying accuracy in relation to the reference direct fetal electrocardiography. The aim of this work was to recognize a source of errors and to develop the method for correction of the indices describing the FHR variability for a given type of fetal monitor. The proposed correction relies upon removing of the constant error component, which has been assigned to an averaging nature of the autocorrelation function. Although the remaining random error component is still not too satisfactory considering the instantaneous values, a significant improvement of reliability of the fetal heart rate variability measurement was confirmed in case of a global one-hour trace assessment.

The aim of the study was to analyse relationships between isopotential areas in sequences of EEG top maps recorded before, during and after seizure activity. Method of the analysis of isopotential areas in sequences of EEG maps consists on the comparison of the changes of values of areas before and after seizure activity and estimation of mutual changes of the areas for the extreme isopotential regions during seizure activity. Results of the study performed on two groups of totally 17 subjects suggest that selected image features like area of given range of potentials and the analysis of the relationships between areas of the isopotential regions before, during and after seizure episode reveal the differences in considered groups of subjects.

Monitoring of a pregnancy course is one of socially important application of biomedical engineering in clinical medicine. In this paper we evaluated a possibility of a nonlinear analysis of an electrohystegraphical signal for assessment of an uterine contractile activity during a pregnancy. This analysis was performed based on a sample entropy statistic. The obtained initial results confirmed that this method could provide clinical useful information for an obstetrical care.

An outline of issues relating to the contemporary application of monoclone antibodies and techniques for obtaining hybrids has been presented in the study. The results of research concerning modifications of Zimmerman’s methods have been presented. They prove a higher efficacy and selectivity of the solution proposed. An electroporation system with changed geometry of electrodes and current parameters has been presented. The procedure is controlled entirely by a microprocessor from the stage of technological parameters’ control in the incubators of initial cells to the creation of a determined mixture of cells in a mixer, then division into portions, pumping the mixture into a hybridisation chamber and, following hybridisation, passing the mixture into separate sections of the incubator. The hybridisation system was modified. A transparent water coat was constructed and connected to a thermostat, on which a transparent hybridisation chamber was installed. Lighting from underneath and gap lighting of the chamber enable continuous observation of suspended cells by means of a microscope lens which is connected by a picture channel to a computer. The software analyses the picture in terms of hybrid selection. The marked cells are planimetrically analysed during the programmed duration. When the morphometric criteria are met, the suspended cells are pumped over to separate sections of the incubator, where selective breeding of hybrids is carried out. The selection of hybrids takes place in electroosmosis gradient under morphometric control of cells in microcapillary systems.

The microscopic images of the cells are very difficult to analyze and to segment. The advanced method of segmentation such as region growing, watershed or snake requires the initialization information about the rough position of the cell body. It is proposed to localize cells in image using a threshold of simplified image. Clustering grey levels in image is proposed to simplify image. The

k

-means clustering method supported by weighting coefficients is chosen to collect all grey tones presented in the background into one cluster and other grey tones into few clusters in such a way that they cover a cell region in microscopic images. The weighting coefficients are used to influence (expand or contract) patterns in microscopic images of living cells. The method was evaluated on the basis of confocal and bright field microscopy images of cells in culture.

The article presents an algorithm and an application to estimate haplotype frequencies from genotype data for unrelated individuals. Presented approach can handle loci with multiple alleles as well as silent (null) alleles. The mathematical model and an expanded Expectation-Maximization algorithm is described. The computer program, called NullHap, available freely at http://staff.elka.pw.edu.pl/ rnowak2/nullhap implements presented ideas. Comparison with known software to estimate haplotypes: Arlequin, PHASE and Haplo-IHP proves the advantage presented method.

Automation of monitoring and analysis of cell culture condition is crucial for fast and reliable optimization of culturing methods e.g. in the regenerative medicine. So the method of automatic cells counting during culture monitoring and analysis of cells population, according to the model of cell culture development, is needed. To solve this general problem several separated problems should be investigated. One of them is the detection of cells which are soon going to divide, the moment when in the new frame two daughter cells could be expected in the place occupied by a single cell in the previous image. A method of localization of cells, which potentially would divide, is proposed in the paper. The mathematical morphology operations are used to detect small, rounded and converged cells which

in potentia

could be mitotic in a single frame from the sequence of images. The proposed method is applied to images of neural stem cells, made with the laser scanner confocal microscope. Results of the experiments are also presented in the paper.

Visualization of protein molecular structures is a very important part of the analysis of protein function and activity. Visualization tools allow to represent graphically the complex construction of proteins and give us an idea of the biological molecules built with hundreds or thousands of atoms linked to each other by covalent bonds. In the chapter we present the most interesting features of our Protein Molecular Viewer (PMV). The PMV is a molecular visualization tool which is used to show protein structures loaded from the well-known Protein Data Bank. With the possibility of loading and presenting protein structures from the PDBML data format the PMV becomes one of a few tools in the world having this unique function.

The current study presents two approaches to the fuzzy support vector machine. The first approach implements the fuzzy support vector machine for solving a two class problem. The second approach employs the fuzzy support vector machine for a multi-class problem. In both cases fuzzy classifiers have been used for genes expression data analysis. The first method has been tested on clinical data acquired at the Silesian Medical University. Then the dataset from Kent Ridge Biomedical Data Set Repository has been used to simulate the performance of the second tool.

This paper reports a comparative study demonstrating what level of predictive performance can be achieved if class prediction is attempted based on features obtained as the top most differently expressed genes from class comparison studies. Several typically used methods of gene ranking in class comparison are considered including Wilcoxon rank test, signal to noise and fold-change method. Predictive performance is estimated for a variety of feature set dimensionalities, this allows to empirically find a classification model yielding best performance for new data. This is used as a measure of predictive performance of feature vectors. Predictive performance is illustrated using publicly available microarray data sets. Results are compared with those using feature selection methods aiming to reduce feature redundancy.

The aim of the study was to test diagnostic potential of a computer-assisted system for identification of neoplastic urothelial nuclei. Presence of neoplastic urothelial nuclei in organic fluid points to neoplastic changes. The system analyzed Feulgen stained cell nuclei obtained with bladder washing technique. Image analysis was carried out by means of a digital image processing system designed by the authors. Features describing nuclei population were measured, then a multistage classifier was constructed to identify positive and negative cases. The principle of the worked out urothelial nuclei analysis on the basis of nuclei size distribution and the basic idea of the case classification were presented. The results obtained in a study of 38 new cases were compared with those obtained with earlier studies. All together 170 cases were analyzed. The results of this new study together with earlier investigated cases yielded ~60% correct classification rate in the control group, while a 86% was obtained among the cancer patients. The predictive value of the positive result of the test based on this method showed to be ~82% and the predictive value of the negative result occurred to be ~75%.

The results shown that this system may be sufficiently well developed to be used successfully in clinical practice.

The paper presents a concept of bioprosthesis control via recognition of user intent on the basis of myopotentials acquired from his body. The contextual recognition of elementary actions is considered and recognition algorithm based on fuzzy inference system with mathematical evidence model is described. Proposed method was experimentally tested on the real data dealing with grasping of different objects limited to the sequences of seven steps of elementary actions.

The Dempster-Shafer theory extended for fuzzy focal elements can be used to build a flexible model of medical diagnosis. Yet, quality of an inferred diagnosis depends on precision of matching knowledge with evidence (patient’s findings). The paper provides definitions of matching precision and suggests methods of the most adequate use of available information about symptoms. The methods are illustrated by an example and tests of an Internet database.

Chronic renal failure is associated with major biochemical and hematological derangements. These changes are often represented as linear functions of creatinine. The aim of the study is to analyze the correlation of hematologic parameters with creatinine. The sample population involved patients with renal insufficiency observed in the Stefan Kardynal Wyszynski Regional Specialists’ Hospital in Lublin (Poland). The method presented here is based on the theory of statistical kernel estimators, which frees it of assumptions in regard to the form of regression function.

The Multiple Classifier Systems are nowadays one of the most promising directions in pattern recognition. There are many methods of decision making by the group of classifiers. The most popular are methods that have their origin in vote methods, where the decision of the common classifier is a combination of simple classifiers decisions. On the other hand there exists a trend of combined classifiers, which are making their decisions basing on the discrimination function, this function is a combination of above-mentioned simple classifier functions. This work presents an attempt to estimate the classifier error, which bases on the combined discrimination function. Obtained from this estimation conclusions will serve to formulate project guidelines for this type of decision-making systems. At the end experimental results of combining algorithms are presented.

The more information is processed in a system the more likely is that some input values are missing. The paper describes (i) a method for managing the incomplete input data in a Mamdani fuzzy system and (ii) discusses the influence of inference interpretation on an efficiency of the fuzzy system operating on incomplete data. Two fuzzy models of missing information are discussed theoretically and then presented on an example of iris data set. Various interpretations of fuzzy rules and various types of membership function are examined in order to find a solution of fuzzy system that is more robust to missing data.

In the paper certain expert system concept for neediness of patients disease entity diagnoses was presented. Problem of knowledge acquisition to knowledge base and conclusion on that knowledge were realized. Testing results of HYBRIDEX expert system were presented as well as the effectiveness of that system usage as a doctor support in setting correct medical diagnose was estimated.

The main purpose of this paper is to deliver appropriate tool to find similarities between texts. The area of interest covers comparing large amount of different texts grouped in various areas of knowledge. Similarity is defined as distance between two texts and as this the measure may be calculated as the set of parameters based on features.

An application of a two-stage classifier to the prognosis of sacroileitis is presented in the paper. The method of classification is based on a decision tree scheme. A

k

-nearest neighbors is applied in pattern recognition task. In this model of classification a fuzzy loss function is used. The efficiency of this algorithm is compared with the algorithm based on zero-one loss function. In this paper also influence of choice of parameter

λ

in selected comparison fuzzy number method on classification results are presented.

Standard methods used for an identification of risk factors are based on logistic regression models. These models disabled to assessment a nonlinearity between a study factors and a disease occurrence. This paper presents an application of generalized additive models for modeling of reproductive risk factors associated with endometrioid cysts. Moreover theoretical similarity and differences between generalized additive models and neural networks was discussed. The obtained results enabled to propose a new etiological aspect for endometrioid cysts.

In this study we examined whether it could be possible to recognize a type of chemical stimuli, given in intermittent cycles, on the basis of observed changes in the breathing pattern in an animal model. Ventilatory responses to three chemical stimulus - normoxic cycles (3-min administration of stimulus/8-min normoxic recovery) in awake rats were investigated. Two types of chemical stimuli were given: (a) gas mixtures of 14% or 9% O

2

in N

2

(i.e. hypoxia), and (b) 5% or 10% CO

2

in O

2

(i.e. hypercapnia), each one in a separate run of the intermittent stimulus. Ventilatory features: respiratory frequency, tidal volume, minute ventilation, inspired and expired times, were used for recognition of ventilatory responses to exposures of the intermittent stimuli. The quality of recognition was evaluated by a probability of misclassification that was estimated experimentally. As a classifier we used the

k

nearest neighbor (

k

-NN) rule that is one of most powerful method offered by the pattern recognition theory. Satisfactory recognition was obtained for recovery periods and stronger stimuli. The best recognition was observed for the intermittent hypercapnia. In conclusion, the approach based on

k

-NN rule has appeared to be useful tool for recognition changes of ventilatory responses to exposures of the intermittent chemical stimuli.

Telesfor system is the result of an initiative undertaken by researchers from Microsoft Innovation Center in Poznan, that attempts to bring a user-friendly and safe system for the support of communication in healthcare. Telesfor is based on ConferenceXP architecture. It enables a document-based collaboration that is enhanced by exchange of text messages and videoconferences between multiple participants of a communication session. It also brings tools for effective organization of a session allowing to address differences in role and competence of participants. In this paper, we present Telesfor system and show how it may be used in telemedical consultations and in distance learning.

Autism is an incurable neurological disease. The main symptoms of this disease are problems with communication and social behaviour. There are difficulties with integration of sensory impressions, too [1]. Nowadays, the only way of solving the problem of autism is rehabilitation, the aim of which is for an autistic person to achieve the best level of functioning in a society.

On account of communication problems of autistic people in their education, it is essential to use methods which allow a teacher to move away to the further plan. This is possible by means of virtual reality methods. This kind of computer program is presented in this paper. The program is designed for rehabilitation of autistic children. The main aim of the program is to show typical human behaviour in daily situations and to make the autistic children familiar with emotions expressed by facial expressions.

Paper describes a computer system enabling interactive online presentation of multimedia Daisy books over the Internet. The system cooperates with the Internet multimedia library computer management system. The main goal of both projects and their execution, is easy and effective access to information for visually impaired people. We focus on new feature of our DaisyReader which allows interactive voice reading of math formulas.

The aim of the work was assessment of stability of tibia fixation realized with the use of double threaded screw. Biomechanical analysis of the tibia – double threaded screw system was carried our for the implant made of two biomaterials used in bone surgery – Cr-Ni-Mo stainless steel and Ti-6Al-4V alloy. Finite element method was applied to calculate displacements, strains and stresses. The obtained results allowed to work out biomechanical characteristics of the analyzed system. These characteristics can be a basis for selection of degree of strain hardening of the applied metallic biomaterial and optimization of the screw’s geometry.

The fundamental purpose of research was determination of biomechanical characteristic of lumbar spine–transpedicular stabilizer system made of stainless steel (Cr-Ni-Mo) and Ti6Al4V alloy. To define biomechanical characteristic of the system finite element method was applied. Geometric models of part spine L3-L4 and stabilizer, was discretised by means of SOLID 95 element. Appropriate boundary conditions imitating phenomena in real system with appropriate accuracy were established. The aim of biomechanical analysis was calculation of displacements and stresses in the vertebras and the stabilizer in a function of the applied loading: 700 N–1600 N. The results of the numerical analysis can be applied to determine a construction features of the stabilizer, and to select mechanical properties of metallic biomaterial. The defined displacements for vertebras L3-L4 show that the proposed type of stabilizer enables correct course of treatment.

The paper presents results of numerical analysis of new form of expandable intramedullary nail (patent no. P382247) used in stabilization of proximal femur in adults. The obtained results can be used to optimize geometry of implants as well as mechanical properties of metallic biomaterial they are to be made of.

The aim of the work was assessment of system for corrective osteotomy of tibia (patent no. P382316). The system consisted of the plate of shape adapted to anatomical curvature of bone and the distance block, assembled together with the plate by means of connective screws. Biomechanical analysis of the tibia – plate system was carried our for the implant made of two biomaterials used in bone surgery – stainless steel and Ti-6Al-4V alloy. Finite element method was applied to calculate displacements, strains and stresses. The obtained results allowed to work out biomechanical characteristics of the analyzed system. These characteristics can be a basis for selection of degree of strain hardening of the applied metallic biomaterial and optimization of the plate’s geometry.

The paper presents the results of kinematic analysis of therapeutic movements of the upper limb, according to PNF method recommendations. Real trajectories of upper limb movements were recorded using the photogrammetric method. The measuring site consisted of a set of 8 digital cameras, two computer workstations, a set of markers, calibrating dice and light sources. On the basis of the recorded images and calculations performed with the use of specialized software, model trajectories of the analyzed movements and values of relative angular translocations and angular velocity in individual joints of the limb were defined.

Influence of mesh density on the results of FEM model analysis of mechanical biocompatibility of dental implants has not been presented yet. Taking advantage of the Ansys v.11 software, carried out was an analysis in the linear elastic range level of stresses in bone tissue surrounding standard osseointegrated implant of complete denture, with a decreasing size of elements (thetragonal type SOLID 187) adjacent to cortical bone/implant interface,respectively 0.5, 0.3 and 0.1. Equivalent Huber-Mises’ stress value in the zone that is exposed to effort the most, and is located close to the edge of implant insertion into the cortex bone significantly increases along with mesh density from app. 60 MPa to 120 MPa for opposite models, because of the lack of convergence of stresses at this singularity point. Increase of mesh density leads to an overestimation of loading stresses values and furthermore to an unjustified increase of pillars’ diameter. At the other hand, too large elements might lead, through an underestimation of loading stress level, to overloading atrophy of bone tissue or to implant loss.

In this contribution, we analyse the dielectric response of a simple RC model composed of single RC circuits (relaxors) with the exponential response. The relaxation time of a single relaxor is defined by the product of resistance and capacitance:

τ

= RC . We present two hypothetical algorithms of generating gaussian input function, one with fluctuation of a density decay, and the other one with normalized density. Both of them enable to obtain a different impedance response of the system. An algorithm of input function creation is crucial to generating the response of the presented model.