Introduced for the first time on July 2008 the WIRELESS ACQUISITION SYSTEM FOR BIOMEDICAL SIGNAL developed by a Biolab team of the Polytechnic University of Turin.

The Technological applications for Sport will be mainly realated to the walking analysis.

The aim of this work is the design and carrying out of an acquisition system characterized by an easy positioning of sensors on the body, low encumbrance and discomfort for the patient, easy data storing and visualization.
The system herein described consists of a wireless network placed on the patient, that links the sensors attached to the body to a central node. This node has two tasks: it manages the information received and sends it to a work station by a USB connection or, if the node is not linked to a work station, stores data on an SD-card. The current version of the system works with four sensors at the same time, and each sensor can process different kinds of signals. In fact it is possible to use a sensor dedicated to EMG acquisition (this kind of sensor acquires EMG signal with a 12 bit resolution and a sample rate of 1 kHz) or, by reducing the sampling frequency, a multipurpose sensor with up to eight input channels.
This system can be used for clinical investigation allowing a real-time visualization of the biomedical signals and the subsequent analysis of data acquired. It can also be applied for long time (i.e., 24 hours) investigation; in this case, data recorded during the entire session are stored on the SD-card. This kind of acquisition net is easier to be placed than a wired net and it is more usable from the patient as well as from the experimenter point of view. Moreover, this net is particularly suitable for analyzing complex and fast movements.

METHODS

The wireless net is obtained using a ZigBee-compliant microchip. ZigBee is a wireless protocol which grants robustness and reliability in the communication, but also an easy realization of the net. The chips used are Ember EM250 and the entire net is developed using the EmberZNet 3.1.0 development kit.
In particular, the development tools used are two: the xIDE programming interface and the InSignt Desktop network evaluation tool. XIDE is a powerful interface for firmware debug and test. This application is used to develop the firmware of both central node (sink) and sensors of the wireless net. This two applications are different: the sink is responsible for the net administration and data transmission to the data management unit, while the sensor is involved in signal conditioning and wireless data transmission to the sink. The InSight Desktop is an interesting tool for the net evaluation. This application visualizes a list of all the data transfer sessions among the nodes of the network and for each data transfer shows the sender ID, the receiver ID, the information sent and the duration of the communication. This tool is really helpful during the test of the maximum data rate which the wireless application can accept without losing stability.

Besides the wireless network, our application needs a data management unit which sends data coming from the net to the PC using a USB communication port or stores these on an SD-card. For this part of the project we used a Cypress microchip PSoC for USB communication: the CY8C24794. The development tool for this device is the PSoC Designer, a user friendly application for the configuration of the digital and analog blocks of the PSoC. The communication with the SD-card is implemented using an SPI master block and reaches a data rate of 3Mb per second. The SD-card is always written following the FAT16 protocol, so data stored can be read from a standard PC.
The USB communication requires writing a driver for this specific application. For these needs Cypress offers a free application, the SuiteUSB.NET, very useful for the setting up of the USB communication. In particular there is a dummy driver for Cypress devices. By modifying this simple file it is possible to obtain a driver for a specific application. This Cypress tool also provides a program written in Visual C# for the management of the input and output data flow to/from the USB port.

RESULTS AND DISCUSSION

The result of our research activity is the implementation of a prototype of wireless system for biomedical signal analysis. At this time, the net is able to receive and store a stream of data with a throughput of approximately 6 kBaud without losing stability. This value depends on the sink communication speed. Considering that each EMG wireless sensor has a sampling rate of 1kHz and a 12-bit resolution, it follows that it requires 1.5 kBaud: the net can then support up to four EMG sensors.
The sink device is placed with the data management unit in a support similar to a USB pen. This pen is the core of our application. It can be self powered using a battery or externally powered by the USB port. Two buttons are located on the surface of this device: the reset and the “allow join” button. The latter is used to link new sensors to the network. In order to disconnect a sensor from the network is sufficient to turn it off.
The sensor devices perform biomedical signal conditioning and sampling and have to ask the sink permission to join the network. We developed two kinds of sensor devices: the EMG sensors (figure 1) and the general purpose sensors.
The EMG sensor is a cylinder  8 mm high with 26 mm diameter and in weight 10 g battery included. Each base of the cylinder is an integrated circuit (figure 2): the upper is dedicated to the wireless communication (figure 2,a) and the digital processing (figure 2,b), while the lower is responsible of the analog conditioning of the EMG signal (figure 2.c). On this circuit has a programmable gain between 250 and 25000 and this value is set during the set-up of the network.
The general purpose sensors can have up to eight input channels. The sampling rate is limited to 100Hz for each channel with 12-bit resolution.
The software of the application is a multi task Visual C# application written to manage real-time signals, to read and to analyze stored acquisitions. During real-time visualization it is possible to set different values of gain for different EMG sensors in order to improve the usage of the range of the A/D converter.

CONCLUSIONS

In conclusion, the application herein presented is a first attempt to realize a completely wireless system for biomedical signal analysis and long term patient monitoring. There are still limitations, such as the maximum number of sensors and the power consumption. To overcome these problems, we are developing a more efficient net structure.

For further details please contact:

Federico De Gioannini – PhD Student

f_de_gioannini@yahoo.it