Hartmut Dickhaus

Classifying biosignals with wavelet networks [a method for noninvasive diagnosis]

In recent years, a particular challenge has arisen in noninvasive medical diagnostic procedures. Because biosignals recorded on the body surface reflect the internal behaviour and status of the organism or its parts, they are ideally suited to provide essential information of these organs to the clinician without any invasive measures. But how are the recorded time courses of the signals to be interpreted with regard to a diagnostic decision? What are the essential features and in what code is the information hidden in the signals? These questions are typical of so-called pattern-recognition tasks. This article reviews pattern recognition as it applies to medical diagnostics and discusses the concept of wavelet networks as a means of biosignal classification. An example is presented in which this approach was used for classifying preprocessed ECG signals to identify patients who were at high-risk of developing ventricular tachycardia (VT).

Automated sleep stage identification system based on time-frequency analysis of a single EEG channel and random forest classifier

In this work, an efficient automated new approach for sleep stage identification based on the new standard of the American academy of sleep medicine (AASM) is presented. The propose approach employs time-frequency analysis and entropy measures for feature extraction from a single electroencephalograph (EEG) channel. Three time-frequency techniques were deployed for the analysis of the EEG signal: Choi-Williams distribution (CWD), continuous wavelet transform (CWT), and Hilbert-Huang Transform (HHT). Polysomnographic recordings from sixteen subjects were used in this study and features were extracted from the time-frequency representation of the EEG signal using Renyis entropy. The classification of the extracted features was done using random forest classifier. The performance of the new approach was tested by evaluating the accuracy and the kappa coefficient for the three time-frequency distributions: CWD, CWT, and HHT. The CWT time-frequency distribution outperformed the other two distributions and showed excellent performance with an accuracy of 0.83 and a kappa coefficient of 0.76.

A deformable digital brain atlas system according to Talairach and Tournoux

Brain atlases are valuable tools which assist neurosurgeons during the planning of an intervention. Since a printed atlas book has several disadvantages-among them the difficulty to map the information onto a patients individual anatomy-we have developed a digital version of the well-established stereotaxic brain atlas of Talairach and Tournoux. Our atlas system is mainly dedicated to assist neurosurgical planning, and its benefits are: (i) a three-dimensional (3D) representation of most brain structures contained in the Talairach atlas; (ii) a nonrigid matching capability which warps the standard atlas anatomy to an individual brain magnetic resonance imaging (MRI) dataset in a few minutes and which is able to take deformations due to tumors into account; (iii) the integration of several sources of neuroanatomical knowledge; (iv) an interface to a navigation system which allows utilization of atlas information intraoperatively. In this paper we outline the algorithm we have developed to achieve 3D surface models of the brain structures. Moreover, we describe the nonrigid matching method which consists of two tasks: firstly, point correspondences between the atlas and the patient are established in an automatic fashion, and secondly these displacement vectors are interpolated using a radial basis function approach to form a continuous transformation function. To generate appropriate target structures for the first of these tasks, we implemented a quick segmentation tool which is capable to segment the cortex and ventricles in less than 5 min. An evaluation shows that our nonrigid approach is more precise than the conventional piecewise linear matching, though it should be further improved for the region around the deep grey nuclei. Summarizing, we developed a Win32 program which permits the convenient and fast application of standardized anatomy to individual brains which potentially contain tumors.

Artefacts in magnetic resonance imaging caused by dental material

Abstract.A common problem in computer tomography (CT) based imaging of the oral cavity is artefacts caused by dental restorations. The aim of this study was to investigate whether magnetic resonance imaging (MRI) of the oral cavity would be less affected than CT by artefacts caused by typical dental restorative alloys. In order to assess the extent of artefact generation, corresponding MRI scans of the same anatomic region with and without dental metal restorations were matched using a stereotactic frame. MRI imaging of the oral and maxillofacial region could be performed without reduction of the image quality by metallic dental restorations made from titanium, gold or amalgam. Dental restorations made from titanium, gold or amalgam did not reduce the image quality of the MRI sequence used in imaging of the oral and maxillofacial region for dental implant planning. In this respect MRI is superior to CT in implant planning.

Single-sweep analysis of event-related potentials by wavelet networks-methodological basis and clinical application

Trial-to-trial variabilities in event-related potentials (ERPs), which are neglected by investigating averaged ERPs, can be important to establish group-specific effects in clinical studies. Single ERP responses have to be analyzed to quantify these variations. In order to overcome the disadvantages of existing single-sweep estimators, the authors have developed a new procedure based on wavelet networks (WNs) and applied this novel approach in a study concerning attention deficit hyperactivity disorder (ADHD) in children. WNs represent signals as a linear combination of wavelet nodes, i.e., components characterized by time-frequency features related to the wavelet transformation. In single-sweep analysis, each wavelet node is restricted to a specific region of the time-frequency plane during the recursive WN training process. This is achieved by means of tapering and bandpass filtering with Gaussian functions which are automatically adapted and closely related to the Morlet basis wavelet. The time course of a single event-related response can be reliably estimated. Furthermore, the WN method automatically provides well-defined parameters for single event-related responses, respectively ERP trial-to-trial variabilities. In a psychophysiological study on ADHD using auditory evoked potentials (AEPs), latency and amplitude parameters extracted from averaged ERPs did not reveal any significant differences between 25 control and 25 ADHD boys. In contrast, interesting group-specific differences could be established by WN single-sweep analysis. In conclusion, WN single-sweep analysis can be recommended as a sensitive tool for clinical ERP studies which should be applied in addition to the investigation of averaged responses.

Time-on-task analysis using wavelet networks in an event-related potential study on attention-deficit hyperactivity disorder

OBJECTIVE The aim of this event-related potential (ERP) study was to test time-on-task analysis at the level of single sweeps in a clinical trial. Since inattentiveness is one of the main symptoms of attention-deficit hyperactivity disorder (ADHD), this child psychiatric disorder was chosen as an exemplary application. METHODS Twenty-four healthy and 24 ADHD boys, aged 9--15 years, performed an auditory selective attention task for about 5 min. ERP single trials were analyzed using wavelet networks. Time-on-task analysis was applied to omission errors, reaction time and slow ERP components (frontal negativity, parietal positivity), represented by a low-frequency wavelet component. RESULTS Both performance and ERP measures showed distinct temporal dynamics. Time-on-task effects were not only linear, but also of higher order and started after less than 1 min. For ADHD children, earlier time-on-task effects, i.e. an earlier increase of omission errors and frontal negativity, resulted. Healthy children could allocate more attentional resources during the course of the experiment. CONCLUSION Time-on-task analysis at the level of single trials revealed phenomena probably reflecting ADHD childrens attentional deficits. Thus, a more differentiated ERP analysis may provide a better understanding of the pathophysiological background in neuropsychiatric disorders.

Classification of Sleep Stages Using Multi-wavelet Time Frequency Entropy and LDA

BACKGROUND The process of automatic sleep stage scoring consists of two major parts: feature extraction and classification. Features are normally extracted from the polysomnographic recordings, mainly electroencephalograph (EEG) signals. The EEG is considered a non-stationary signal which increases the complexity of the detection of different waves in it. OBJECTIVES This work presents a new technique for automatic sleep stage scoring based on employing continuous wavelet transform (CWT) and linear discriminant analysis (LDA) using different mother wavelets to detect different waves embedded in the EEG signal. METHODS The use of different mother wavelets increases the ability to detect waves in the EEG signal. The extracted features were formed based on CWT time frequency entropy using three mother wavelets, and the classification was performed using the linear discriminant analysis. Thirty-two data sets from the MIT-BIH database were used to evaluate the performance of the proposed method. RESULTS Features of a single EEG signal were extracted successfully based on the time frequency entropy using the continuous wavelet transform with three mother wavelets. The proposed method has shown to outperform the classification based on a CWT using a single mother wavelet. The accuracy was found to be 0.84, while the kappa coefficient was 0.78. CONCLUSIONS This work has shown that wavelet time frequency entropy provides a powerful tool for feature extraction for the non-stationary EEG signal; the accuracy of the classification procedure improved when using multiple wavelets compared to the use of single wavelet time frequency entropy.

Bridging the scales

MOTIVATION Biological reality can in silico only be comprehensively represented in multi-scaled models. To this end, cell behavioural models addressing the multi-cellular level have to be semantically linked with mechanistic molecular models. These requirements have to be met by flexible software workflows solving the issues of different time scales, inter-model variable referencing and flexible sub-model embedding. RESULTS We developed a novel software workflow (EPISIM) for the semantic integration of Systems Biology Markup Language (SBML)-based quantitative models in multi-scaled tissue models and simulations. This workflow allows to import and access SBML-based models. SBML model species, reactions and parameters are semantically integrated in cell behavioural models (CBM) represented by graphical process diagrams. By this, cellular states like proliferation and differentiation can be flexibly linked to gene-regulatory or biochemical reaction networks. For a multi-scale agent-based tissue simulation executable code is automatically generated where different time scales of imported SBML models and CBM have been mapped. We demonstrate the capabilities of the novel software workflow by integrating Tysons cell cycle model in our model of human epidermal tissue homeostasis. Finally, we show the semantic interplay of the different biological scales during tissue simulation. AVAILABILITY The EPISIM platform is available as binary executables for Windows, Linux and Mac OS X at http://www.tiga.uni-hd.de. Supplementary data are available at http://www.tiga.uni-hd.de/supplements/SemSBMLIntegration.html. CONTACT niels.grabe@bioquant.uni-heidelberg.de.

Radial, spiral and reverberating waves of spreading depolarization occur in the gyrencephalic brain

OBJECTIVES The detection of the hemodynamic and propagation patterns of spreading depolarizations (SDs) in the gyrencephalic brain using intrinsic optical signal imaging (IOS). METHODS The convexity of the brain surface was surgically exposed in fourteen male swine. Within the boundaries of this window, brains were immersed and preconditioned with an elevated K(+) concentration (7 mmol/l) in the standard Ringer lactate solution for 30-40 min. SDs were triggered using 3-5 μl of 1 mol/l KCl solution. Changes in tissue absorbency or reflection were registered with a CCD camera at a wavelength of 564 nm (14 nm FWHM), which was mounted 25 cm above the exposed cortex. Additional monitoring by electrocorticography and laser-Doppler was used in a subset of animals (n=7) to validate the detection of SD. RESULTS Of 198 SDs quantified in all of the experiments, 187 SDs appeared as radial waves that developed semi-planar fronts. The morphology was affected by the surface of the gyri, the sulci and the pial vessels. Other SD patterns such as spirals and reverberating waves, which have not been described before in gyrencephalic brains, were also observed. Diffusion gradients created in the cortex surface (i.e., KCl concentrations), sulci, vessels and SD-SD interactions make the gyrencephalic brain prone to the appearance of irregular SD waves. CONCLUSION The gyrencephalic brain is capable of irregular SD propagation patterns. The irregularities of the gyrencephalic brain cortex may promote the presence of re-entrance waves, such as spirals and reverberating waves.

Modeling multi-cellular behavior in epidermal tissue homeostasis via finite state machines in multi-agent systems

MOTIVATION For the efficient application of multi-agent systems to spatial and functional modeling of tissues flexible and intuitive modeling tools are needed, which allow the graphical specification of cellular behavior in a tissue context without presuming specialized programming skills. RESULTS We developed a graphical modeling system for multi-agent based simulation of tissue homeostasis. An editor allows the intuitive and hierarchically structured specification of cellular behavior. The models are then automatically compiled into highly efficient source code and dynamically linked to an interactive graphical simulation environment. The system allows the quantitative analysis of the morphological and functional tissue properties emerging from the cell behavioral model. We demonstrate the relevance of the approach using a recently published model of epidermal homeostasis as well as a series of cell-cycle models. AVAILABILITY The complete software is available in binary executables for MS-Windows and Linux at tiga.uni-hd.de.