2017 22nd International Conference on Digital Signal Processing (DSP), 2017
The paper presents specific methods of processing of multimodal data recorded during physical act... more The paper presents specific methods of processing of multimodal data recorded during physical activities by depth MS Kinect cameras, thermal imaging cameras and heart rate sensors. All video data and heart rate signals used in the present study were recorded in the home environment. The proposed methodology includes the detection of the chest breathing area for breathing motion analysis used by the MS Kinect. For the thermal image processing the static and dynamic selection of regions of interests was performed in associated sets of images to find time evolution of respiratory signals and their temperature changes. Signal de-noising by finite impulse filters is applied both for breathing and heart rate data. Correlation analysis is used in the data processing stage to find the time relation between individual physiological variables. Results include relations between signals acquired during physical activities and they show how simple sensors can be used to increase the accuracy of standard diagnostical tools in biomedicine as well.
2020 3rd International Conference on Emerging Trends in Electrical, Electronic and Communications Engineering (ELECOM), 2020
The paper is devoted to the analysis of multichannel biomedical signals acquired in the sleep lab... more The paper is devoted to the analysis of multichannel biomedical signals acquired in the sleep laboratory. The whole data set includes 29 overnight records of electroencephalogram (EEG), electrooculogram (EOG), electrocardiogram (ECG), sound and movement data observed with a sampling frequency of 200 Hz, and breathing records (Flow) acquired with a sampling frequency of 10 Hz, among others. The methodology used for their processing includes their filtering, feature extraction and classification using both standard and deep learning classification methods as well. The goal of the paper is (i) in presentation of the deep learning machine learning in the frequency domain without any specification of features, (ii) in comparison with results obtained by the classical approach based on the two layer neural network model and initial specification of signal features, and (iii) in comparison of separation of the Wake sleep stage from the REM, NonRem1, NonREM2, and NonREM3 sleep stages using the deep learning method. The best separation accuracy of 92.12 % (with the loss value 0.19) was achieved for the separation of the Wake and NonREM3 stages for a single EEG channel and data segments 30s long. Results suggest that the deep learning strategy can help with sleep stages classification in the clinical environment
ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2019
In this article, detection of sleep apnea or hypopnea events is addressed using a single channel ... more In this article, detection of sleep apnea or hypopnea events is addressed using a single channel electrocardiography (ECG) signal by analysis of respiratory extracted modulation. First, R peaks are detected from ECG signal. Then, a time-series with the amplitude (height) and timing of R peaks representing respiratory-induced amplitude modulation is constructed. This signal is resampled evenly at 4Hz. Synchrosqueezed wavelet transform (SSWT) together with an iterative time-frequency ridge estimation is applied to provide a robust estimation of instantaneous respiratory frequency and detect the regions with/without sleep apnea/hypopnea events. Signal reconstruction using inverse synchrosqueezed wavelet transform (ISSWT) has been performed. The appeared peaks can identify and measure the duration of apnea/hypopnea events.
2009 17th European Signal Processing Conference, 2009
The paper presents the possibility of using a sliding window for image feature extraction in orde... more The paper presents the possibility of using a sliding window for image feature extraction in order to identify image regions of interest. The study includes the comparison of feature extraction methods both in the space and frequency domains using the discrete Fourier transform and the discrete wavelet transform to achieve reliable classification results for a given application. The compactness of feature clusters is evaluated exploiting a proposed numerical criterion. In case of real image data, the clusters compactness can often be improved by employing a chosen smoothing method on the raw data. In this paper, the procedure of smoothing, feature extraction and classification is applied to microscopic images of aluminum alloys in order to isolate regions of similar properties and to study their relationship. To achieve this goal the sliding window classification results are combined and isolated misclassified subregions repaired. The proportion of misclassified regions is then used...
The paper describes basic operations associated with the use of the distributed computing toolbox... more The paper describes basic operations associated with the use of the distributed computing toolbox and its application for processing of extensive and complex mathematical problems using the computer network and the set of computers for parallel processing of separate components of the whole algorithm.
2017 22nd International Conference on Digital Signal Processing (DSP), 2017
The paper presents a new algorithm for adaptive classification of sleep stages using multimodal d... more The paper presents a new algorithm for adaptive classification of sleep stages using multimodal data recorded in the sleep laboratory during overnight polysomnography records. The proposed method includes the learning process applied for the set of individuals with their sleep stages classified by an experienced neurologist. Features evaluated for time windows 30 s long and selected multimodal signals are used for construction and optimization of the proposed two-layer neural network model. Resulting computational system based upon breathing EEG and EOG features is used for analysis of new individuals to detect their sleep stages. Results include classification accuracy higher than 80% and 90% for Wake and REM stages, respectively. The proposed method can adaptively modify model coefficients to detect sleep stages and sleeping disorders using man-machine interaction.
Texture analysis and classification of image components belong to common problems of the interdis... more Texture analysis and classification of image components belong to common problems of the interdisciplinary area of digital signal and image processing. The paper is devoted to the pattern matrix construction using features evaluated by the discrete Fourier transform (DFT) or the discrete wavelet transform (DWT) using the relative power in selected frequency bands or scale levels, respectively. Image features are then used to recognize groups of similar pattern vectors by self-organizing neural networks forming a mathematical tool for cluster analysis. Further classification methods including the decision tree, support vector machine, nearest neighbour method and neural networks are then applied for construction of specific models and evaluation of their accuracy and cross validation errors. The proposed algorithm is applied for analysis of given microscopic images representing wax structures covering breathing openings on leaves affected by environmental pollution in different locat...
The paper presents a new method for enhancement of orthodontic images using the combination of tw... more The paper presents a new method for enhancement of orthodontic images using the combination of two-dimensional data with different type of illumination and their processing using digital de-nosing and gradient image enhancement in the preliminary stage. The region growing method and the appropriate seed selection is then studied as an alternative to the distance and watershed transforms for their segmentation. The main goal of the paper is in (i) the presentation of mathematical methods for image analysis in the orthodontic treatment and (ii) presentation of the region growing method use for image regions classification. Resulting algorithms are used for gradient edge detection and image components enhancement in orthodontics. Key–Words: Biomedical image processing, object illumination, digital filters, gradient methods, image segmentation, region growing method
The paper is devoted to analysis of orthodontic images and to new mathematical methods for evalua... more The paper is devoted to analysis of orthodontic images and to new mathematical methods for evaluation of their characteristics. The main goal of the paper is in analysis of three-dimensional objects to enable numerical evaluation of measures important for the study of the appropriate treatment after dental operations. Methods presented include (i) computer analysis of a single image based upon its de-noising followed by a thresholding and image components detection and (ii) presentation of tools for the three dimensional modelling using the double camera system. Proposed algorithms allow semi-automatic evaluation of measures between selected objects.
Electromyography (EMG) represents a method used for the acquisition of electrical activity produc... more Electromyography (EMG) represents a method used for the acquisition of electrical activity produced by skeletal muscles. The following analysis based upon the digital signal processing methods can be used to find relation between their neurological activation and separate motor units firing with the typical frequency of 7-20 Hz. Signals obtained can be used for analysis of biomechanics of the human and for the detection of medical abnormalities including those of muscles (myopathy) and neurons (neuropathy). The paper presents the use of wavelet transform for data analysis for the selected decomposition level and the threshold value. Results include (i) the description of the proposed graphical user interface and (ii) comparison of results for healthy and neuropathic patients.
2017 22nd International Conference on Digital Signal Processing (DSP), 2017
The paper presents specific methods of processing of multimodal data recorded during physical act... more The paper presents specific methods of processing of multimodal data recorded during physical activities by depth MS Kinect cameras, thermal imaging cameras and heart rate sensors. All video data and heart rate signals used in the present study were recorded in the home environment. The proposed methodology includes the detection of the chest breathing area for breathing motion analysis used by the MS Kinect. For the thermal image processing the static and dynamic selection of regions of interests was performed in associated sets of images to find time evolution of respiratory signals and their temperature changes. Signal de-noising by finite impulse filters is applied both for breathing and heart rate data. Correlation analysis is used in the data processing stage to find the time relation between individual physiological variables. Results include relations between signals acquired during physical activities and they show how simple sensors can be used to increase the accuracy of standard diagnostical tools in biomedicine as well.
2020 3rd International Conference on Emerging Trends in Electrical, Electronic and Communications Engineering (ELECOM), 2020
The paper is devoted to the analysis of multichannel biomedical signals acquired in the sleep lab... more The paper is devoted to the analysis of multichannel biomedical signals acquired in the sleep laboratory. The whole data set includes 29 overnight records of electroencephalogram (EEG), electrooculogram (EOG), electrocardiogram (ECG), sound and movement data observed with a sampling frequency of 200 Hz, and breathing records (Flow) acquired with a sampling frequency of 10 Hz, among others. The methodology used for their processing includes their filtering, feature extraction and classification using both standard and deep learning classification methods as well. The goal of the paper is (i) in presentation of the deep learning machine learning in the frequency domain without any specification of features, (ii) in comparison with results obtained by the classical approach based on the two layer neural network model and initial specification of signal features, and (iii) in comparison of separation of the Wake sleep stage from the REM, NonRem1, NonREM2, and NonREM3 sleep stages using the deep learning method. The best separation accuracy of 92.12 % (with the loss value 0.19) was achieved for the separation of the Wake and NonREM3 stages for a single EEG channel and data segments 30s long. Results suggest that the deep learning strategy can help with sleep stages classification in the clinical environment
ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2019
In this article, detection of sleep apnea or hypopnea events is addressed using a single channel ... more In this article, detection of sleep apnea or hypopnea events is addressed using a single channel electrocardiography (ECG) signal by analysis of respiratory extracted modulation. First, R peaks are detected from ECG signal. Then, a time-series with the amplitude (height) and timing of R peaks representing respiratory-induced amplitude modulation is constructed. This signal is resampled evenly at 4Hz. Synchrosqueezed wavelet transform (SSWT) together with an iterative time-frequency ridge estimation is applied to provide a robust estimation of instantaneous respiratory frequency and detect the regions with/without sleep apnea/hypopnea events. Signal reconstruction using inverse synchrosqueezed wavelet transform (ISSWT) has been performed. The appeared peaks can identify and measure the duration of apnea/hypopnea events.
2009 17th European Signal Processing Conference, 2009
The paper presents the possibility of using a sliding window for image feature extraction in orde... more The paper presents the possibility of using a sliding window for image feature extraction in order to identify image regions of interest. The study includes the comparison of feature extraction methods both in the space and frequency domains using the discrete Fourier transform and the discrete wavelet transform to achieve reliable classification results for a given application. The compactness of feature clusters is evaluated exploiting a proposed numerical criterion. In case of real image data, the clusters compactness can often be improved by employing a chosen smoothing method on the raw data. In this paper, the procedure of smoothing, feature extraction and classification is applied to microscopic images of aluminum alloys in order to isolate regions of similar properties and to study their relationship. To achieve this goal the sliding window classification results are combined and isolated misclassified subregions repaired. The proportion of misclassified regions is then used...
The paper describes basic operations associated with the use of the distributed computing toolbox... more The paper describes basic operations associated with the use of the distributed computing toolbox and its application for processing of extensive and complex mathematical problems using the computer network and the set of computers for parallel processing of separate components of the whole algorithm.
2017 22nd International Conference on Digital Signal Processing (DSP), 2017
The paper presents a new algorithm for adaptive classification of sleep stages using multimodal d... more The paper presents a new algorithm for adaptive classification of sleep stages using multimodal data recorded in the sleep laboratory during overnight polysomnography records. The proposed method includes the learning process applied for the set of individuals with their sleep stages classified by an experienced neurologist. Features evaluated for time windows 30 s long and selected multimodal signals are used for construction and optimization of the proposed two-layer neural network model. Resulting computational system based upon breathing EEG and EOG features is used for analysis of new individuals to detect their sleep stages. Results include classification accuracy higher than 80% and 90% for Wake and REM stages, respectively. The proposed method can adaptively modify model coefficients to detect sleep stages and sleeping disorders using man-machine interaction.
Texture analysis and classification of image components belong to common problems of the interdis... more Texture analysis and classification of image components belong to common problems of the interdisciplinary area of digital signal and image processing. The paper is devoted to the pattern matrix construction using features evaluated by the discrete Fourier transform (DFT) or the discrete wavelet transform (DWT) using the relative power in selected frequency bands or scale levels, respectively. Image features are then used to recognize groups of similar pattern vectors by self-organizing neural networks forming a mathematical tool for cluster analysis. Further classification methods including the decision tree, support vector machine, nearest neighbour method and neural networks are then applied for construction of specific models and evaluation of their accuracy and cross validation errors. The proposed algorithm is applied for analysis of given microscopic images representing wax structures covering breathing openings on leaves affected by environmental pollution in different locat...
The paper presents a new method for enhancement of orthodontic images using the combination of tw... more The paper presents a new method for enhancement of orthodontic images using the combination of two-dimensional data with different type of illumination and their processing using digital de-nosing and gradient image enhancement in the preliminary stage. The region growing method and the appropriate seed selection is then studied as an alternative to the distance and watershed transforms for their segmentation. The main goal of the paper is in (i) the presentation of mathematical methods for image analysis in the orthodontic treatment and (ii) presentation of the region growing method use for image regions classification. Resulting algorithms are used for gradient edge detection and image components enhancement in orthodontics. Key–Words: Biomedical image processing, object illumination, digital filters, gradient methods, image segmentation, region growing method
The paper is devoted to analysis of orthodontic images and to new mathematical methods for evalua... more The paper is devoted to analysis of orthodontic images and to new mathematical methods for evaluation of their characteristics. The main goal of the paper is in analysis of three-dimensional objects to enable numerical evaluation of measures important for the study of the appropriate treatment after dental operations. Methods presented include (i) computer analysis of a single image based upon its de-noising followed by a thresholding and image components detection and (ii) presentation of tools for the three dimensional modelling using the double camera system. Proposed algorithms allow semi-automatic evaluation of measures between selected objects.
Electromyography (EMG) represents a method used for the acquisition of electrical activity produc... more Electromyography (EMG) represents a method used for the acquisition of electrical activity produced by skeletal muscles. The following analysis based upon the digital signal processing methods can be used to find relation between their neurological activation and separate motor units firing with the typical frequency of 7-20 Hz. Signals obtained can be used for analysis of biomechanics of the human and for the detection of medical abnormalities including those of muscles (myopathy) and neurons (neuropathy). The paper presents the use of wavelet transform for data analysis for the selected decomposition level and the threshold value. Results include (i) the description of the proposed graphical user interface and (ii) comparison of results for healthy and neuropathic patients.
Uploads
Papers by Ales Prochazka