Click to add title

Download Report

Transcript Click to add title

Saint-Petersburg State Polytechnical University
Research work
Mathematical modeling and streamed data
processing
Supervisor: Ph.D., associate professor of CS & SE dept. Bolsunovskaya M.V.
1
Directions of scientific research

The system analysis, computer simulation and information processing technologies;

Development and research of new methods for object processing, localization and
recognition;

Algorithms and software tools development for data processing and visualization;

Three-dimensional modeling and prototyping;

Technologies and software for non-destructive testing.
2
Applications of Research and Development
Technical systems and technological complexes :
• Systems for complex technical objects management;
• Distributed systems and industrial technological complexes;
• The hardware and software systems of industrial automation;
• Quality management system.
Information Systems :
• Medical information systems;
• Geographic Information Systems;
• System of non-destructive testing in the field of biotechnology, medicine
and materials;
• High-performance storage and data processing systems.
3
Research and development results

R&D «Development of information system for quality assessment and production cycle
of medical devices "Vascular valve homografts "" ( with a help of program to support
innovative projects in science and technology «START-2004», 2004-2011);

R&D «Development and research of a prototype complex for remote dynamic pulse
oximetry control» (with a help of program to support innovative projects in science and
technology «START-2006», 2006-2008);

R&D " Algorithms development for tomographic images for the new generation of CT
scanners with low-dose irradiation (under the international cooperation agreement with
Philips Corporation», 2012);

R&D «Computer information technologies and technical means of control and
modeling in large systems» (commissioned by the Ministry of Education and Science of
the Russian Federation, 2012).
4
Software complex for the production support and clinical use of medical
devices «Vascular valve homografts»
 The purpose of development - commercial production of
medical devices for performing cardiac surgery for children
and adults with heart defects.
 During the 7 years of development we’ve created a system to
support all stages of the production process - from the stage of
assessing the quality of raw materials to the stage of evaluation
of the results of clinical use.
5
Software complex for the production support and clinical use of medical devices
«Vascular valve homografts»
The following have been developed:

Methods for time series regression analysis, classification (modified algorithm of the
decision tree, Naive Bayes algorithm) and clustering (k-means algorithm, the
expectation maximization);

Quantitative and qualitative analysis of the digital signal and image processing,
obtained by the control equipment (microscopic analysis, CT, MRI).

Methods of data mining to identify patterns in the results of the product usage, based on
the statistical data processing algorithms (correlation analysis, regression analysis,
Kaplan-Meier estimations).
6
Informational complex for material digital image processing
Developer:
Leksashov A.V.
7
Digital image processing tool “SPM image processing”
Developers:
Leksashov A.V., Bolsunovskaya M.V., Timofeeva T.A., Tronina E.I.
8
Complex for remote dynamic pulse oximetry control
Features
Allows to:

implement dynamic monitoring of oxygen saturation
and pulse rate of patients of all ages on the stage of
intensive care and rehabilitation;

evaluate the cardiovascular and respiratory systems;

transfer data between system components using
wireless communication channels within the
department of medical institutions.






Patent:
«System for monitoring body status» (№59389
27.12.2006).
Developers: Odoevsky A.S.,
Fomin A.M.,
Rudakov A.S., Timofeeva T.A.
Institution: Saint-Petersburg children hospital №1


Complex specifications:
Indicators: heart rate, saturation;
Wireless Bluetooth connection;
Up to 6 pulse oximeters at a time ;
Autonomous power supply;
Operator
workplace
:
Laptop or Desktop PC;
Statistical
analysis
(mean,
deviation,
correlation, etc.);
DB Interbase/Firebird, long-term storage;
Report export (MS Word).
9
Complex for remote dynamic pulse oximetry control
Real-time tracking
Individually customizable settings and alerting
Сохранение информации в БД Interbase/Firebird
Developers: Odoevsky A.S.,
Fomin A.M.,
Rudakov A.S., Timofeeva T.A.
Institution: Saint-Petersburg children hospital №1
10
Complex for remote dynamic pulse oximetry control
Statistical processing of the stored data:
Developers: Odoevsky A.S.,
Fomin A.M.,
Rudakov A.S., Timofeeva T.A.
Institution: Saint-Petersburg children hospital №1
11
Complex for remote dynamic pulse oximetry control
Microsoft office word report format
Developers: Odoevsky A.S.,
Fomin A.M.,
Rudakov A.S., Timofeeva T.A.
Institution: Saint-Petersburg children hospital №1
12
Software for multiscale model time series prediction
 Wavelet-decomposition;
 Decomposition parameters (number of levels, the type of
wavelet);
 Visualization of decomposition in the format of the spectrogram
or the time series;
 Predictive model assignment for each level of decomposition;
 The calculation of the forecast errors;
 Step-based simulation mechanism, allowing to return to any of
the iterations passed;
 GUI;
 Manual or file-based data input;
 Support for different export formats(bmp, png, xls и plain text);
 Automation of the entire simulation cycle.
13
Software for multiscale model time series prediction
The original
series
Optimal decomposition:


I
Energy-entropy
Cross-correlation
Data classification:



The fractal
dimension
Ljung-Box test
Estimation of
the variance,
the fluctuations
Levels of
representation
II
Models of
levels
Models:
Model n
Model 3
Model 2
Model 1


III


Time series
forecast


Holt-Winters
The harmonic
regression
Method of the
least squares
Spline
interpolation
ARIMA (AIC)
ETS
14
Software for multiscale model time series prediction
Development tools: Python + R + Qt
15
Areas of co-operation with industry
• Fundamental and applied research in the field of mathematical modeling and highperformance computing systems for imaging and non-destructive testing.
• The techniques and technology of design and implementation of new systems, numerical
calculation methods, simulation, visualization and computer graphics hardware and
software solutions for industry and research.
• Algorithms, methodologies and software tools for data storage and processing of stream
data.
16