Gateway for:

Member Countries

Methods and algorithms of digital processing of biomedical signals

#B-705


Development of Biomedical Signals Processing Methods, Algorithms and Software for Diagnostics of Cardiovascular and Speech Production Organs Diseases

Tech Area / Field

  • MED-DID/Diagnostics & Devices/Medicine
  • INF-SIG/Sensors and Signal Processing/Information and Communications
  • INF-SOF/Software/Information and Communications

Status
8 Project completed

Registration date
23.03.2001

Completion date
27.01.2006

Senior Project Manager
Rybakova T A

Leading Institute
Belarussian State University, Belarus, Minsk

Supporting institutes

  • Republican Science-Practical Center Cardiology, Belarus, Minsk\nRepublican Clinic of Hearing, Voice and Speech Pathology, Belarus, Minsk

Collaborators

  • National Technical University of Athens / Department of Electrical and Computer Engineering, Greece, Athens\nUniversity of Tokyo / Department of Biomedical Engineering, Japan, Tokyo\nCNRS / Université de Rennes 1 / Laboratoire Traitement du Signal et de l'Image INSERM, France, Rennes\nTechnical University of Aachen / Institute for Statistics, Germany, Aachen\nUniversity of Illinois At Urbana-Champaign / Department of General Engineering, USA, IL, Urbana\nNational Technical University of Athens / Department of Electrical Engineering and Computer Science, Greece, Athens

Project summary

Project purpose:

Development of methods, algorithms and software for digital biomedical signal processing for diagnostics of cardiovascular and speech production organs diseases.

Task statement.

Methods and tools for digital processing of signals of different nature are intensively developed during last decades. Substantial results at this sphere were achieved by developing and practical application of new methods and algorithms, among which it is necessary to mark out the following ones: wavelet transforms (M.Akay. "Time, Frequency and Wavelets in Biomedical Engineering". IEEE Press, Piscataway, N.J., 1997.), hidden Markov models (L.Rabiner, B.H.Juang. Fundamentals of Speech Recognition. Prentice-Hall, 1993.), genetic and neural-network algorithms (David Goldberg etc. Genetic Programming 1998. Academic Press. Morgan Kaufmann 1998).

The tasks of biomedical signal processing belong to the sphere where the application of new methods of digital signal processing is able to provide quick and important results. According to data of the expert committee of World health-protection organization the leading place among disease and mortality rates pertains to cardiovascular diseases, in particular, to the most widespread one - the arterial hypertension. Basing on the results of international multi-center researches it was established that regular treatment of arterial hypertension decreases the appearance probability of different complications like brain isulite at 42%, and the mortality from ischemia at 16%. In spite of successes at arterial hypertension treatment the adequate control of arterial pressure can be established only for 25-48% of patients at the moment. Thereupon the early disease diagnostics, choice of optimal inpidual treatment scheme and complication preventive measures make up the actual problem which has the bright social- economic aspect in addition to the medical one.

At the disease origin stage the complex (nonlinear and non-stationary) combination of biometric values changes can be observed. Thereupon within the project frameworks first it is first time proposed a new approach to biomedical signals analysis based on wavelet transforms, autoregressive hidden Markov models and genetic algorithms for early man disease diagnostics.

Proposed project influence to the progress at the development area.

Traditional methods of functional diagnostics are highly tailored ones. Most of modern diagnostics systems provide monitoring of biophysical and biochemical parameters only. That is why the development of methods and algorithms for digital processing of biomedical signals and systems for diagnostic solutions basing on large amount of biophysical parameters under the conditions of high a priory uncertainty is an important and actual science-technical problem. The solution of this task makes up the context of the project, and it will let to found new diagnostics methods for cardiovascular and speech production organs diseases.

Project participants.

The scientific team includes two research groups from BSU, which consist of experts at the sphere of systems and tools for missile target tracing at anti-missile means, at the sphere of aiming and target designation systems and automatic self-direction systems. Within the project framework specialists from BSU will develop methods, algorithms and software for digital biomedical signals processing. Specialists from the Republican science-practical center “Cardiology” will develop methods of early diagnostics of cardiovascular system diseases and carry out clinical tests of the developed software. Main role of specialists from the Republican clinic of hearing, voice and speech pathology is in the development and tests of speech production organs diagnostics methods.

Expected results.

Within the framework of the proposed project it will be solved 6 tasks and the following main results will be achieved:

1. The feature vector forming algorithms for biomedical signals will be developed and investigated based on wavelet transforms.

2. The methods and algorithms of hidden Markov models building for biomedical signal processing tasks will be developed.

3. The genetic algorithms for biomedical signal processing will be developed.

4. The robust dicriminant analysis methods for autoregressive sequences and wavelet coefficients will be developed.

5. The vocal tract disease identification system will be developed based on hidden Markov models and genetic algorithms.

6. The method and PC-software complex for early diagnostics of arterial hypertension and complications preventive measures will be developed based on wavelet transforms and genetic algorithms.

The project results application.

Scientific results, achieved during the project fulfillment, can be used:

– for hardware-software means creation for disease diagnostics and pathology exposure of speech production organs, for the development of practical methods for early diagnostics of speech production organs diseases;


– for development of methods and hardware-software means of coronary pathology early diagnostics for patients with arterial hypertension, and for these diseases treatment scheme optimizing;
– for model and software means design for forecast of arterial hypertension evolution and complications depending on disease progress variants and medical therapy efficiency.

As commercial and industry meanings of the project it is possible to mark out: the development of the PC-based system for speech production organs disease identification, and the development of hardware-software complex for early diagnostics of arterial hypertension and complications preventive measures. The design of such systems is an important social and economical task since, for example, in the USA more then 60 mln of people have different forms of arterial hypertension, for Belarus this figure is equal to 3 mln, which is 33% of population. Opportune diagnostics of this disease is able to decrease costs for its treatment and social insurance.

Compliance of project activity with ISTC aims:

– the project realization will let to redirect the activity of the BSU research group, who were employed at the sphere of target designation, aiming and automatic self-direction systems design, systems and tools for missile target tracing at anti-missile means, to peaceful problem solving, concerned with disease diagnostic complex development;


– the participating in the project will help the specialists to expand international scientific links and let the further development of research results within the framework of partner projects;
– financial support of the project will let to develop new effective technologies for disease diagnostics and patient health monitoring taking into account the multifactor influence of the Chernobyl A-plant damage consequences;
– scientific team, participating at the project, in future can compose the basis of a scientific and technical firm, functioning at the market of multi-purpose medical diagnostic systems.

Data about work volume.

The project duration is 36 months. During the project fulfillment it will be solved 6 tasks, pided into 16 stages.

The role of foreign collaborators.

The cooperation with project collaborators will promote the fruitful scientific information exchange within the project researches framework, the objective estimates and reviewing of scientific results, the active participating in development and introduction of practical results. The following collaboration forms are supposed:

– S.Ueno, Head of the URSI committee “Elecromagnetic waves in biology and medicine”, Professor, Dept. of Biomedical Engineering, Grad. School of Medicine, University of Tokyo, Japan.

Planned participating form: scientific information exchange within the project research framework, estimates and reviews of scientific results, achieved during the project fulfillment, help in project support.

– A.Constantinides, Professor, Head of the scientific group on communications and signals, Imperial College, Great Britain.

Planned participating form: scientific information exchange within the project research framework, estimates and reviews of scientific results, achieved during the project fulfillment, participating in science and practical project results discussing, help in project support.

– J.L.Coatrieux, Professor, Director of the Laboratorie Traitement du Signal et de l’Image, Universit de Rennes, France.

Planned participating form: scientific information exchange within the project research framework, participating in science and practical project results duscussing, help in project support.

– D.Goldberg, Professor, Director of the Dept. of General Engineering, University of Illinois at Urbana-Champaign, USA.

Planned participating form: scientific information exchange within the project research framework, participating in science and practical project results discussing, estimates and reviews of scientific results, achieved during the project fulfillment.

– N.Uzunoglu, Professor, Dept. of Electrical Engineering and Computer Science, National Technical University of Athens, Greece.

Planned participating form: scientific information exchange within the project research framework, estimates and reviews of scientific results, achieved during the project fulfillment, help in project support.

– H. Bock, Professor, Institute For Statistics Technical University of Aachen, Germany.

Planned participating form: scientific information exchange within the project research framework, participating in science and practical project results discussing, help in project support.

Technical approach and methodology.

For biomedical signals modeling it is proposed to use hidden Markov models with observed values, generated by the appropriate autoregressive model. For the first time it will be developed algorithms for multi-layer hidden Markov model training, based on the parameter re-estimation procedure with genetic algorithms usage. The hidden evolution method is proposed for reducing of biomedical signal feature space and variability influence decreasing. It is proposed the multi-stream disease identification method, when identification is carried out by independent separate model characters, such approach provides high identification model accuracy.

For diagnostics of arterial hypertension and its complications the methods of parametric dicriminant analysis will be used. In order to classify signals it is proposed to apply the discriminant analysis of autoregressive sequences, for the coefficients estimation of such kind of sequences the robust methods, stable to “outliers” and distortions, will be developed, and algorithms for informative characters selection will be designed. For signal classification there will be jointly used methods of wavelet transforms and discriminant analysis.

Project participants have done a large amount of preliminary work, there were more then 30 papers published in leading world journals and international conferences proceedings. Project participants have a good experience at the sphere of applied software development. This fact lets to be sure that the purpose and tasks of the project will be fulfilled in time and on high science-technical level.


Back