The article presents the new approach to a computer users verification. The research concerns an analysis of user’s continuous activity related to a keyboard used while working with various software. This type of analysis constitutes a type of free-text analysis. The presented method is based on the analysis of users activity while working with particular computer software (e.g. text editors, utilities). A method of computer user profiling is proposed and an attempt to intrusion detection based on k-NN classifier is performed. The obtained results show that the introduced method can be used in the intrusion detection and monitoring systems. Such systems are especially needed in medical facilities where sensitive data are processed.

The paper proposes possible improvements in signature recognition approach based on window method. The analysis focuses on a stage of window preprocessing using fuzzy sets in order to choose significant ranges of each signature. Proposed extension allows the solution to improve in two areas. First of all minimizing a number of processed windows significantly reduces computation time. Secondly, filtered signatures with valuable information about significant ranges allow the system to recognize signatures of a poor or good quality. Developed method of signature quality assessment can be used in any signature recognition system, regardless of used method of analysis. Merging the information about signature quality and choosing only important signature ranges should also improve the overall detection results, however, more examinations are needed to confirm this statement.

In the work authors applied speech recognition techniques to find disfluent events. The recognition system based on the Hidden Markov Model Toolkit was built and tested. The set of context dependent HMM models was trained and used to locate speech disturbances. Authors were not concentrated on specific disfluency type but tried to find any extraneous sounds in a speech signal. Patients read prepared sentences, the system recognized them and then results were compared to manual transcriptions. It allowed the system to be more robust and enabled to find all disfluencies types appearing at word boundaries. Such system can by utilized in many ways, for example like a "preprocessor"that finds strange sounds in a speech to be analyzed or classified by other algorithms later, to evaluate or track therapy process of stuttering people, to evaluate speech fluency by ´normal´speakers, etc.

This paper presents a new multilayer ensemble of classifiers for users verification who use computer keyboard. The special keyboard extracts the key pressure and latency between keyboard keys pressed during password entered. When user is typing password the system creates a pattern based on time and key pressure. For users verification group of classifiers have been proposed. It allows to obtain the higher accuracy level compared to alternative techniques. The efficiency of the proposed method has been confirmed in the experiments carried out.

The article is related to the computer systems security and to the problem of detecting the masqueraders, intruders who pretend to be legitimate users of computer systems, in particular. The research concerns computer user verification based on analysis of the mouse activity in computer system. The article presents an improvement of user verification method by introducing a new method of user profiling. The new user profiling method is based on the reduced number of features without the loss of information is introduced. The presented method of user profiling allows to simplify and speed up a user’s activity data analysis and is not causing the deterioration of intrusion detection. Additionally, a method of aggregating mouse activity basic events into a higher level events is described. This article presents the preliminary research and conclusions.

The Recent rapid growth of the information technology aid people to improve the performance in processing their job activities. These changes, in turn, force them to carry out their job activities using computers in the workplace; that results in the increase of somatic and/or mental stresses. In this study, we conducted a preliminary experiment to estimate an impact of web-browsing on human mind and body. Two types of web-browsing tasks, which are a 18-minuetes of continuous web-browsing and an intermittent web-browsing (first 45 second of each 1-minutes interval), were given to the subjects (10 healthy male aged from 20-23) with within-subjects experimental design. With regard to physiological measures, two prominent stress biomarkers, salivary immunoglobulin A (IgA) and cortisol, were employed. Comparing among the task conditions, relatively larger and long-lasting increase of IgA and smaller decrease of cortisol was observed with an intermittent web-browsing despite there was no difference in the psychological state. These results illustrate the difficulty of estimating the mental workload caused by web-browsing and the importance of employing physiological indices.

The paper presents a personal identification method based on lips photographs. This method uses a new approach to the extraction and classification of characteristic features of the mouth from photographs. It eliminates the drawbacks that occur during the acquisition of lip print images with the use of the forensic method that requires special tools. Geometrical dimensions of the entire mouth as well as of the upper and lower lips were adopted as the features, on the basis of which the verification is performed. An ensemble classifier was used for the classification of the features obtained. The effectiveness of the classifier has been verified experimentally.

In this paper a new method for lip print recognition is proposed. The proposed approach is based on Fuzzy c-Means clustering of the characteristics features of lip prints. First, the Hough transform is applied for the recognition of the characteristic features within lip prints, then Fuzzy c-Means clustering is performed to cluster those features. The proposed algorithm applies the results of clustering to find an unknown image withing the collected repository of lip prints. Instead of comparing all pairs of individual characteristic features, the proposed algorithm uses the representatives of clusters for the comparison of images. The advantage of using the proposed method is its increased tolerance to the noise in data and thus the increased efficiency of the recognition. The effectiveness of presented method has been verified experimentally using real-world images. The results are satisfactory and suggest the possibility of using the method in forensic identification systems.

This work presents the multiscaled version of modified census features in graphical objects detection with AdaBoost cascade training algorithm. Several experiments with face detector training process demonstrate better performance of such features over ordinal census and Haar-like approaches. The possibilities to join multiscaled census and Haar features in single hybrid cascade of strong classifiers are also elaborated and tested. The high resolution example images were used in detector training process.

The medical data and its classification have to be treated in particular way. The data should not be modified or altered, because this could lead to false decisions. Most state-of-the-art classifiers are using random factors to produce higher overall accuracy of diagnosis, however the stability of classification can vary significantly. Medical support systems should be trustworthy and reliable, therefore this paper proposes fusion of multiple classifiers based on artificial Neural Network (ANN). The structure selection of ANN is performed using granular paradigm, where granulation level is defined by ANN complexity. The classification results are merged using voting procedure. Accuracy of the proposed solution was compared with state-of-the-art classifiers using real medical data coming from two medical datasets. Finally, some remarks and further research directions have been discussed.

In this work the topic of applying clustering as a knowledge extraction method from real-world data is discussed. The authors propose hierarchical clustering method and visualization technique for knowledge base representation in the context of medical knowledge bases for which data mining techniques are successfully employed and may resolve different problems. What is more, the authors analyze the impact of different clustering parameters on the result of searching through such a structure. Particular attention was also given to the problem of cluster visualization. Authors review selected, two-dimensional approaches, stating their advantages and drawbacks in the context of representing complex cluster structures.

Paper describes a novel modification to a well known kNN algorithm, which enables using it for medical data, which often is a class-imbalanced data with randomly missing values. Paper presents the modified algorithm details, experiment setup, results obtained on a cross validated classification of a benchmark database with randomly removed values (missing data) and records (class imbalance), and their comparison with results of the state of the art classification algorithms.

The paper presents the dynamic ensemble selection based on the analysis of the decision profiles. These profiles are obtained from a posteriori probability functions returned from the base classifiers during the training process. Presented in the paper dynamic ensemble selection algorithms are dedicated to the binary classification task. In order to verify these algorithms, a number of experiments have been carried out on several medical data sets. The proposed dynamic ensemble selection is experimentally compared against the ensemble with the sum fusion method. As base classifiers we used the pool of homogeneous classifiers. The obtained results are promising because we could improve the classification accuracy of the ensemble classifier.

In the presented work two variants of the fuzzy clustering approach dedicated for determining the antecedents of the rules of the fuzzy rule-based classifier were presented. The main idea consists in adding additional prototypes (’prototypes in between’) to the ones previously obtained using the fuzzy c-means method (ordinary prototypes). The ’prototypes in between’ are determined using pairs of the ordinary prototypes, and the algorithm based on distances and densities finding such pairs was proposed. The classification accuracy obtained applying the presented clustering approaches was verified using six benchmark datasets and compared with two reference methods.

Pattern classification systems play an important role in medical decision support. They allow to automatize and speed-up the data analysis process, while being able to handle complex and massive amounts of information and discover new knowledge. However, their quality is based on the classification models built, which require a training set. In supervised classification we must supply class labels to each training sample, which is usually done by domain experts or some automatic systems. As both of these approaches cannot be deemed as flawless, there is a chance that the dataset is corrupted by class noise. In such a situation, class labels are wrongly assigned to objects, which may negatively affect the classifier training process and impair the classification performance. In this contribution, we analyze the usefulness of existing tools to deal with class noise, known as noise filtering methods, in the context of medical pattern classification. The experiments carried out on several real-world medical datasets prove the importance of noise filtering as a pre-processing step and its beneficial influence on the obtained classification accuracy.

The paper presents initial research on method, which improves precise indoor localization and steering of autonomous mobile devices that can be used for medical applications like: patient’s state monitoring, medicine distribution or environmental data collection before medical intervention (in case of biohazard or fire). The localization of object is based on optical codes, which are modified to be easily identified from distance in low light. Multiple codes modification was tested to find optimal ones. The visual recognition system is using Hough transform and Canny edge detection to read values from code. The novelty of the proposed method is reading values directly from image, without scaling and rotation. Moreover, the steering algorithm for identified device is proposed. It takes distance and decision uncertainty under consideration. The proposed method was verified against state-of-the-art optical codes in real-world indoor environment. Finally, the further research directions are discussed.

In this paper a method is introduced which enables automatic detection of parathyroid hyperplasia and parathyroid adenoma on the basis of immunohistochemical angiogenesis markers expression in micrographs. The proposed method uses digital image processing techniques and classification algorithms to detect diseased tissue. The disease detection is performed by classification of normalized color intensity histograms. Accuracy of this method was evaluated by using micrographs of parathyroid tissue sections obtained from patients that have undertaken surgery due to primary hyperparathyroidism. Use of different color models, various classifiers, and immunohistochemical markers was considered during the experiments. The experimental results show that the introduced method enables accurate detection of parathyroid disease. The most promising results were obtained for k-nearest neighbor and neural network classifiers.

Modern medical imaging techniques produce huge volume of data from stack of images generated in a single examination. To compress them several volumetric compression techniques have been proposed. Performance of these compression schemes can be improved further by considering the anatomical symmetry present in medical images and incorporating the characteristics of human visual system. In this paper a volumetric medical image compression algorithm is presented in which perceptual model is integrated with a symmetry based lossless scheme. Symmetry based lossless and perceptually lossless algorithms were evaluated on a set of three dimensional medical images. Experimental results show that symmetry based perceptually lossless coder gives an average of 8.47% improvement in bit per pixel without any perceivable degradation in visual quality against the lossless scheme.

The number and shape of cells in endothelium layer is highly correlated with the quality of vision. Therefore, its precise and automatic description plays an important role in medicine. This work presents several aspects of image processing of endothelium layer acquired by specular microscope. The comparison of cell selection accuracy is discussed using two different approaches to solve this problem: convolution filtering methods, and snake-based method. Moreover, for verification results generated by dedicated software, supplied with the microscope, were utilized. Next, the precise segmentation method is applied to improve the segmentation. The results are inspected visually, but also CV, H, and CVSL parameters, used in medicine, are calculated. The research concludes that general visual outcomes achieved by all segmentation approaches give similar results, however deep insight into cell outline position reveals some differences, which were partially removed after precise segmentation application. The analysis of parameter values show high stability of CV and CVSL parameters.

Glioma detection and classification is an critical step to diagnose and select the correct treatment for the brain tumours. There has been advances in glioma research and Magnetic Resonance Imaging (MRI) is the most accurate non-invasive medical tool to localize and analyse brain cancer. The scientific global community has been organizing challenges of open data analysis to push forward automatic algorithms to tackle this task. In this paper we analyse part of such challenge data, the Multimodal Brain Tumor Image Segmentation Benchmark (BRATS), with novel algorithms using partial learning to test an active learning methodology and tensor-based image modelling methods to deal with the fusion of the multimodal MRI data into one space. A Random Forest classifier is used for pixel classification. Our results show an error rates of 0.011 up to 0.057 for intra-subject classification. These results are promising compared to other studies. We plan to extend this method to use more than 3 MRI modalities and present a full active learning approach.

Automated techniques for measuring elasticity parameters of cells enable development of new diagnosis methods. An important elasticity parameter is the Young’s modulus (YM), which has been effectively used to characterize different cell properties, e. g., platelet activation, locomotion, differentiation, and aging. This paper deals with the problem of automated determination of cells YM based on the force-distance curves obtained from atomic force microscope. During experiments, the YM of cells was determined by using contact point detection and curve fitting algorithms. Experimental results were compared for two theoretical models of indentation: Hertz model, and Sneddon model. The results show that single indentation model allows a satisfactory accuracy to be obtained only for a subset of the force-distance curves. The most appropriate model for a given curve can be selected based on the fitting error analysis.

In this paper we present OvaExpert, an intelligent system for ovarian tumor diagnosis. We give an overview of its features and main design assumptions. As a theoretical framework the system uses fuzzy set theory and other soft computing techniques. This makes it possible to handle uncertainty and incompleteness of the data which is an unique feature of developed system. The main advantage of OvaExpert is its modular architecture which allows seamless extension of system capabilities. Two diagnostic modules are described in the paper along with examples. First module is based on aggregation of existing prognostic models for ovarian tumor. Second, on novel concept of Interval–Valued Fuzzy Classifier which is able to operate under data incompleteness and uncertainty.

The paper presents the preliminary study of the utilization of a low resolution ECG signal analysis. The analysis was performed on the signals obtained from a hand-held ECG monitor usually used in primary health care. The aim. The main aim of the study was a registration of series of data by volunteers within couple of months and determination of signal quality and main ECG parameters as follows: Q, R, S waves, QRS duration as well as the end of PQ and the beginning of ST segment. Additionally, the heart rate variability was determined. Materials and methods. The data was registered by 12 volunteers aged from 35 to 55. The ECG tests were carried out for 7 months. The sample rate of the signal was 100 Hz. To determine the ECG parameters the signal processing and statistical methods was used. Results. The sensitivity of the following ECG parameters were: R wave detection - 99,2 %, Q wave detection - 99,1 %, S wave detection - 99,0 %, QRS duration - 99,2 % respectively.

This study is aimed at evaluation of the capability to indicate the preterm delivery risk analysing the features extracted from signals of electrical uterine activity. Free access database was used with signals acquired in two groups of pregnant women who delivered at term and preterm. Signal features comprised classical time domain and spectral parameters of contractile activity, as well as the sample entropy. Their mean values were calculated over all contraction episodes detected in each record and their statistical significance for separating the two groups of recordings was provided. Influence of electrodes location, band-pass filter settings and gestation week was investigated. The obtained results showed that a spectral parameter – the median frequency was the most promising indicator of the preterm delivery risk.

Monitoring of uterine contractile activity enables to control the progress of labour. Automated detection of contractions is to be an integral part of the signal analysis implemented in computeraided fetal surveillance system. Evaluation of efficiency of three algorithms for automated detection of uterine contractions in the signal of uterine mechanical activity is presented. These algorithms are based generally on analysis of the frequency distribution of signal values. The reference data in form of beginning and end of contraction episodes were obtained from human expert. Obtained results showed high efficiency of the algorithms tested where the best one ensured the sensitivity and positive predictive value equal to 92.2 and 97.2, respectively.

The article presents the results of research in controlling the mobile application with the EEG signals and eye blinking. Authors proposed a prototype solution of a brain-computer interface that can be used by people with total motor impairment to control chosen mobile application on their mobile phone. There was a NeuroSky MindWave Mobile device used during experiments. Two software tools for mobile devices were specially implemented. First one helps to analyse the EEG signals and recognize eye blinks, second one - interprets them and executes assigned actions. Different configurations of settings were used during the studies. They included: single blink or double blink, level of focus, period of focus. Experiments results show that a man equipped with a personal EEG sensor and eye blinking detector can remotely touchless use mobile applications installed on smartphones or tablets.

A number of methods of the qualitative assessment of fetal heart rate (FHR) signals are based on supervised learning. The classification methods based on the supervised learning require a set of training recordings accompanied by the reference interpretation. In the real data collections the class of signals related to fetal distress is usually under-represented. Too small percentage of distress patterns adversely affects the effectiveness of the automated evaluation of the fetal state. The paper presents a method of equalizing the class sizes based on the reference assessment of the fetal state with the fuzzy analysis of the newborn attributes. The supervised learning with increased number of the FHR signals, which are characterized by the highest rate of the fuzzy inference leads to significant increase of the efficacy of the qualitative assessment of the fetal state using the Lagrangian support vector machine.

A number of algorithms for estimating the so called fetal heart rate baseline was proposed so far. However, there is no reference pattern enabling their objective evaluation, and thus no methodology of comparing the competing algorithms still exists. In this paper we propose a method for evaluation of automatically determined baseline in reference to a group of experts, basing on ten separate groups of signals comprising typical patterns observed in the fetal heart rate. For the purpose of quantitative assessment of the estimated baseline a new synthetic inconsistency coefficient is presented. The proposed methodology was applied to evaluate ten well-known algorithms. We believe that the method will be a valuable tool for assessment of the existing algorithms, as well as for developing new ones.

The aim of the study was a continuation of previously provided research on the tablets usability in the rehabilitation of old patients with cognitive disabilities after a brain stroke. First, the analysis of the pilot research results was provided. It gave a hope for possible rationalization of the rehabilitation process after proper improvement of the proposed mobile solution. Prototype examples of mobile rehabilitation applications were extended and integrated in the RehaMob - a single mobile solution. The mobile application includes six different types of tasks, which can be useful for stimulation and rehabilitation of language functions, short-term memory, etc. The RehaMob was made available again to the therapists in the rehabilitation center. Finally, the mobile application is used with good results in the therapy of patients with cognitive disabilities. The mobile solution is useful support for traditional rehabilitation process, but its influence to the rehabilitation progress is difficult to assess, because the RehaMob can be used only as additional tool.

In this study, we address the problem of medical diagnosis by applying Fuzzy Cognitive Map (FCM). A distinctive feature of the FCM is its ability to simulate the development of the disease in time. By this simulation, it is possible to predict the severity of the disease by having future knowledge on current medical investigations. For the first time in this paper, we construct an FCM-based classifier dedicated solely to perform medical diagnosis. To learn the FCM, we use an evolutionary algorithm explicitly specifying the newly designed fitness function. Real, publicly available medical data are applied for the validation and evaluation of the proposed approach.