Use of Mobile Embedded Devices for Processing of

Use of Mobile Embedded Devices for Processing of Biomedical Signals Dalibor Janckulik1, Leona Motalova1 and Ondrej Krejcar2 1 dept. of Measurement and Control Technical University of Ostrava, Faculty of Electrical Engineering and Computer Science Ostrava, Czech Republic {Dalibor.Janckulik, Leona.Motalova}@vsb.cz 2 dept. of Measurement and Control Centre for Applied Cybernetics Technical University of Ostrava, Faculty of Electrical Engineering and Computer Science Ostrava, Czech Republic [email protected]

Abstract — Current platform of mobile embedded devices provide a possibility to use them as a monitoring stations of biomedical signals of patients or sportsmen. New platforms are based on windows mobile operating systems from windows CE family. The CPU power is able to process more amounts of biomedical data in Real Time than any before. We are developing an intelligent monitoring system to process an ECG signal from patients. However a problem with Real Time processing of a 12 channels ECG from remote ECG device by Bluetooth still exists here. The problem is based in part of packet parsing. Two possible solutions are discussed. After processing of data a huge need of a presentation of data grown here. A Windows Presentation Foundation (WPF) solution was found as a one possible solution which was also tested.

critical values which could be the sign of worse medical condition of a patient. In the moment of crossing the border of monitored bio-signals values predefined by doctor, system will inform responsible medical staff and provides all information which could help to determine the cause and seriousness of the problem.

Keywords — Embedded, Processing, ECG; Visualization; Biotelemetry

I.

INTRODUCTION

Aim of the platform for patients’ bio-parameters monitoring is to offer a solution providing services to help and make full health care more efficient without limitations for specific country. Physicians and other medical staff will not be forced to make difficult and manual work including unending paperwork, but they will be able to focus on the patients and their problems. All data will be accessible almost anytime anywhere through special applications designated for portable devices web browser or desktop clients and any changes will be made immediately at disposal to medical staff based on the security clearance. Physicians will have immediate access to the patient’s newest results of accomplished examinations. In the case that the ambulance have to go to some accident, rescue team can due to portable devices send information about patient health condition directly to hospital where responsible doctors and staff will have information needed to execute immediate operation without delaying by preparation of necessary equipment. All bio-signals data are stored and automatically analyzed by neuronal network. System can evaluate presence of

Fig. 1 Architecture of Guardian II platform.

The basic idea is to create a system that controls important information about the state of a wheelchair-bound person (monitoring of ECG and pulse in early phases, then other optional values like temperature or oxidation of blood etc.), his situation in time and place (GPS) and an axis tilt of his body or wheelchair (2axis accelerometer). Values are measured with the existing equipment, which communicates with the module for processing via Bluetooth wireless communication technology. Most of the data (according to heftiness) is processed directly in PDA or Embedded equipment to a form that is acceptable for simple visualization. Two variants are possible in case of embedded equipment – with visualization and without visualization (entity with/without LCD display). Data is continually sent by GPRS or WiFi network to a server, where they are processed and evaluated in detail. Processing and evaluating on the

server consists of - receiving data, saving data to data storage, visualization in an advanced form (possibility to recur to the older graph, zoom on a histogram (graph with historical trend), copying from the graphs, printing graphs), automatic evaluation of the critical states with the help of advanced technologies (algorithms) that use Artificial intelligence to notify the operator about the critical state and its archiving. Application in PDA, Embedded equipment is comfortable, with minimum time - the first configuration, but also configuration after downfall of application. The problem we would like to describe is concerned from the processing of ECG data in Real Time. ECG data are the most complex in compare to each other previously mentioned. The problem was found in processing of 12 channels ECG from BT ECG device to database. Our current software application on .NET platform is unable to process these data in real time to provide an online visualization neither on desktop nor on mobile device. We found only two possibilities to surpass this problem. First is in SQL server procedure, second solution is in use of embedded microcontrollers unit HCS08 for data preprocessing. These phenomena will be discussed in chapter 3. II.

DEVELOPED PARTS OF PLATFORM

Complete proposition of solution and implementation of the platform for patient’s biotelemetry as it was described in previous chapter requires determination and teamwork. Every single part of the architecture has to be designed for easy application and connectivity without user extra effort, but user must be able to use given solution easily and effectively. Crucial parts of whole architecture are network servers, database servers and client applications. Due to these crucial parts the development is focused particularly on proposition and implementation of mobile and desktop client application, database structure and some other important web services. A. Server Parts In order to run a server, an operating system supporting IIS (Internet Information Server) is needed. IIS allow to users to connect to the web server by the HTTP protocol. The web service transfers data between the server and PDA/Embedded devices. Web service also read the data, sends acknowledgments, and stores the data in the database. The service is built upon ASP.NET 2.0 technology. The SOAP protocol is used for the transport of XML data. Methods that devices communicating with the web service can use include: • receiving measured data, • receiving patient data, • deleting a patient, • patient data sending. To observe measured data effectively, visualization is needed. A type of graph as used in professional solutions is

an ideal solution. To achieve this in a server application, a freeware Zed Graph library can be used. For data analysis, neural nets are a convenient solution. However, there are problems in the automatic detection of critical states. Every person has a specific ECG pattern. The Neural net has to learn to distinguish critical states of each patient separately. Important part of Guardian is central database. There are stored all data of medical staff and patients. Data of patients include different records such as diagnosis, treatment progress or data which are results of measuring by small portable devices designated to home care. These data represent the greatest problem, because amount of these data rapidly increase with increasing amount of patients. Due to this fact database servers are very loaded. B. Mobile Parts The main part of the system is an Embedded or PDA device. The difference in applications for measurement units is the possibility to visualize the measured data in both Realtime Graph and Historical Trend Graph, which can be omitted on an embedded device. PDA is a much better choice for Personal Healthcare, where the patient is already healthy and needs to review his condition. Embedded devices can be designed for one user, with the option to use an external display used for settings or with the possibility of usage in extreme conditions. Devices based on PDA type have a several limitations such as low CPU performance, low battery life or small display, which is possible to solve by embedded version of such mobile clients. We created a special windows mobile based embedded device. During the development process the several problems occurred. One of them and the most important was the need of a new operation system creation for our special architectural and device needs. We used the Microsoft PlatformBuilder for Windows CE 4.2 tools. The created operation system based on standard windows mobile has several drivers which we need to operate with communication devices and measurement devices. Table 1 Mobile Devices with LCD 480x800 pixels, GSM, WiFi, BT. Mobile device

HTC Touch HD HTC Touch HD2 HTC Touch Diamond 2 Samsung Omnia II

OS WM

Display

6.1 Pro LCD 3,8”

528

SPB Benchmark Index 553

6.5 Pro LCD 4,3”

1000

779

6.5 Pro LCD 3,2”

528

520

6.5 Pro

800

565

LED 3,7”

CPU [Mhz]

As measurement device is possible to connect several device with Bluetooth communication possibility. In our application we use an ECG Measurement Unit (3channels ECG Corbelt or 12 channels BlueECG) through a virtual serial port using wireless Bluetooth technology. Measured data are stored on a SD Memory Card as a database of MS

SQL Server 2008 Mobile Edition. The performance of available devices seems insufficient for sequential access [Table 1]; parsing of incoming packets is heavily timeconsuming. Pseudo paralleling is strongly required. A newer operating system (Windows Mobile 6) must be used to allow the processing of data from a professional EKG due to thread count limitations. C. Parsing of Packets To make an ECG visualisation the measured data are needed at the beginning. The measurement is made on bipolar ECG Corbelt and 12 channels BlueECG. The amount needed to transfer from source device through a Bluetooth is in Table 2. You can compare the increased data transfer speed in case of 12 channels ECG to 1 500 bytes per second. These data amount is very small; on the other hand the data are going as packets, so the processing is needed before the real data can be accessed. Table 2 BT ECG Device -> Mobile Device measurement.

ECG device 3 Channels 12 Channels

Packet Size [Bytes] 100 300

Speed [Packets/s] 3x 5x

Transfer Speed [kB/s] 0,3 1,5

Table 3 BT ECG device measurement. ECG device BT – Mobile device – Server – Visualization BT – Mobile device – Server – Visualization BT – Mobile device – Server – DB Visualization BT – MCU - Mobile device – Server Visualization

Platform .NET Framework C++ SQL Server procedure MCU HCS08

Problem Memory overflow Memory overflow -

-

Real Time Impossible Impossible

Fig. 2 Measurement chain: ECG Bluetooth device – mobile device – aserver – Visualization

D. Database tables and stored procedures The RAW data table contains full size packets received from an ECG device. Only the packets with measured data are stored to database. Those packets must contain in part of packet number bytes with a value of 0x0724 [Table 4]. Table 4 RAW data table

Soft RT (2 sec deadline) Hard RT

This process (called “parsing”) take an unacceptable time in case of mobile device to process the data in Real Time [Table 3]. Same problem is growing on desktop PC, where the C# or managed C++ is used, when we want use universal solution(classes for parsing packet data from more other devices). In both cases the Memory Overflow is reached. The only possible way we found is in use of SQL procedure which is executed on SQL server. When the data (packets) are stored in table the procedure is call to execute and provide RAW data. In such case the data are ready to userconsumer application until 2 second deadline, so the Soft Real Time is possible to use [Fig 2]. Other solution is in using unmanaged C++, C# or C# in .NET MicroFramework. If we use C# with hardly written source code specifically to parsing BlueECG device with bitwise operations, the packet would be parsed in 10 miliseconds. This time is acceptable for us, but the visualization is the weakest part of the parser-vizualization system. This weakness will be solved on desktop pc, trough drawing over WPF in GPU.

The table with parsed data [Table 5] contains decimal values. Column „I“ contains data from bipolar ECG; column „II“ with „I“ contains data from 6-channel ECG. The 12channel ECG fills after parsing columns „I II V1 V2 V3 V4 V5 V6“. Table 5 DATA TABLE OF PARSED DATA

Packet parsing performed group stored procedures, which pulls data from a table of non-parsed data and stores them in the table of parsed data. Package procedure has the same structure as the method of solution in C #. E. Visualizazion WPF (Windows Presentation Foundation) provide up to date possibilities to visualize ECG data on desktop PC. We create a WPF application to provide a full scale of graphic features to user. WPF technology runs directly on GPU possibilities (Graphic Processing Unit) which is founded on modern graphics cards. This fact is key parameter for speed of data presentation of screen. CPU has more time to compute others tasks (e.g. ECG data analyses by neural network tasks). WPF has a more design possibilities in compare to classical Windows Forms including 3D animation, pattern changes of whatever elements etc.. WPF application allows viewing an ECG characteristic of measured patient in Real Time, selection of patient from database and view of historical graphs. Next figure shows an example of bipolar ECG characteristic in WPF application.

Fig. 3 Real Time visualization of bipolar ECG on desktop device in WPF application.

III.

CONCLUSIONS

All developed platforms were tested during these extreme tests with high credibility of measured data for physicians. The measuring device (bipolar Corbelt ECG and 12 channels BlueECG) was tested in extreme conditions in a cryogen room in spa Teplice nad Becvou (Czech Republic) (-136°C). The reached experimental data will be used by physicians to make a set of recommendation for cardiac patient who are healing in this Cryogenic chamber. In such case of patients the time of recovery can be shortness by tens of percent. The Real Time measurement and visualization was reached in case of WPF usage (chapter 3.1). As the final improvement in the future, the application would have some special

algorithm, which could recognize any symptoms of the QRS curve, and make the job for the doctors much easier.

ACKNOWLEDGMENT This research has been carried out under the financial support of the research grants “Centre for Applied Cybernetics“, Ministry of Education of the Czech Republic under Project 1M0567.

REFERENCES [1]

Penhaker, M., Cerny, M., “Sensitivity Analysis in Sensor HomeCare Implementation” In 13TH International Conference on Biomedical Engineering, Vols 1-3 Book Series: IFMBE Proceedings, pp. 979983, December 3 – 6, Volume: 23 Issue: 1-3 [2] Machacek, Z, Srovnal, V., “Automated system for data measuring and analyses from embedded systems”, In 7th WSEAS International Conference on Automatic Control, Modeling and Simulation, Mar 1315, pp. 43-48, Prague, Czech Republic (2005) [3] Dondio, P., Longo, L., Barrett, S., “A translation mechanism for recommendations”, In IFIP, Volume 268, IFIPTM 2008/Joint iTrust and PST Conference on Privacy, Trust Management and Security Trondheim, Nórway, Jun 18-20, 2008, pp. 87-102. (2008) [4] Krejcar, O., Janckulik, D., Motalova, L., “Complex Biomedical System with Mobile Clients”. In The World Congress on Medical Physics and Biomedical Engineering 2009, WC 2009, September 0712, 2009 Munich, Germany. IFMBE Proceedings, Vol. 25/5. O. Dössel, W. C. Schlegel, (Eds.). Springer, Heidelberg. (2009) [5] Cerny, M., Penhaker, M., “Biotelemetry”, In 14th Nordic-Baltic Conference an Biomedical Engineering and Medical Physics, IFMBE Proceedings, Vol 20, Jun 16-20, pp. 405-408, Riga, Latvia (2008) [6] Krejcar, O., Janckulik, D., Motalova, L., Musil, K., Penhaker, M., Real Time ECG Measurement and Visualization on Mobile Monitoring Stations of Biotelemetric System. In The 2nd Asian Conference on Intelligent Information and Database Systems, ACIIDS 2010, March 24-26, 2010 Hue City, Vietnam. Springer, Advances in Intelligent Information and Database Systems, Studies in Computational Intelligence, SCI Vol. 283. pp. 67-78. N.T. Nguyen et al. (Eds.). Springer-Verlag, Berlin, Heidelberg. [7] Krejcar, O., Janckulik, D., Motalova, L., Complex Biomedical System with Biotelemetric Monitoring of Life Functions. In Proceedings of the IEEE Eurocon 2009, May 18-23, 2009, St. Petersburg, Russia. pp. 138-141. DOI 10.1109/EURCON.2009.5167618 [8] Cerny M, Martinak L, Penhaker M, et al. Design and Implementation of Textile Sensors for Biotelemetry Applications In konference proceedings 14th Nordic-Baltic Conference an Biomedical Engineering and Medical Physics, JUN 16-20, 2008 Riga, LATVIA, Volume: 20 Pages: 194-197 , 2008 [9] Penhaker, M., Cerny, M., “The Circadian Cycle Monitoring“ Conference Information: 5th International Summer School and Symposium on Medical Devices and Biosensors, JUN 01-03, 2008 Hong Kong, PEOPLES R CHINA, Pages: 41-43, 2008 [10] Liou, CY., Cheng, WC., “Manifold construction by local neighborhood preservation”, In Springer LNCS, Volume 4985, 14th International Conference on Neural Information Processing (ICONIP 2007) Kitakyushu, Japan, Non 13-16, 2007, pp. 683-692. (2007