Validating wireless EMG results

[Elizabeth Parks]

Hello,
Our lab has implemented a new Delsys wireless EMG system used in conjunction with Vicon motion capture software. We are in need of a procedure to validate the clinical EMG results, test the accuracy of signal acquisition etc.

Does anyone have a similar system in place or a working procedure that would be applicable?
Thanks,
Elizabeth

[Edmund Cramp]

Re: Validating wireless EMG results

The validation of an EMG system with a 3D motion capture equipment is a fairly simply process from a technical point of view, irrespective of the type of system that you use. There are basically two questions to answer:

1. Is the data from the EMG system collected synchronously with the 3D data?
2. Does the EMG data recorded by the 3D system accurately reproduce the signal applied to the input of the EMG system?

One simple test will answer both questions if you apply a 1ms pulse signal simultaneously to one of the EMG system inputs, and directly to an ADC input on the 3D motion capture system. Recording and examining these two signals will answer both questions.

1. The delay between the signal recorded at the ADC input, and the signal recorded from the EMG system, will tell you the EMG system latency - the time that the EMG signal takes to get from the EMG input (i.e. the EMG signal recorded from the subject) until it is detected by the motion capture system ADC. This delay must be less than the frame rate of the 3D system to ensure that the recorded EMG information is synchronized with the recorded 3D and force plate information - less than 16ms (0.016 sec) if the 3D frame rate is 60Hz, and less than 8ms (0.008 seconds) if the 3D frame rate is 120Hz.

The first attached picture (whisper_good.jpg) shows the results of a typical test - the upper trace (green) is the signal applied directly to the ADC, while the delayed lower trace (blue) is the signal applied to the ADC after passing through the EMG system. The delay introduced by the EMG system is measured at 1.16ms (0.0016 seconds) in this example, assuring accurate synchronization between the 3D system and the EMG data on the subject.

2. The second question - accuracy of the EMG signal reproduction - is answered by comparing the shape of the input pulse (green trace) to the shape of the output pulse (blue trace). Ideally the output signal will closely match the input signal however because the EMG system frequency response is not infinite there will be some rounding of the shape of the pulse after passing through the EMG system as shown in the whisper_good.png illustration is a typical example of this, displaying a low latency and minimal distortion of the input pulse.

The second attached picture (whisper_bad.jpg) shows a poor EMG system response with a large EMG latency - in this picture you can see that the second pulse, after passing through the EMG system, is not only considerably delayed (16ms or 0.016 seconds) but also "rings" which introduces additional frequency components into the recorded signal that do not exist in the EMG signal on the subject.

The data shown in the two illustrations was generated using a Whisper EMG simulator (Roessingh Research and Development) and recorded at the ADC inputs using a digital oscilloscope (Dataq Instruments).

In addition to the purely functional tests described above, if you are using a wireless EMG system you should also test the operational range of the system and its ability to operate in the presence of radio-frequency interference since both these attributes will affect the quality of the recorded EMG data.

[Philip Davenport]

Re: Validating wireless EMG results

We had a similar problem here in Birmingham. We the same set up as your lab, with a Vicon motion capture system and a Delsys Trigno EMG box.

We’ve recently completed a testing device that uses a Labview VI as a signal generator to send repeatable signals to 8 sensors at a time. We can compare this response to the sent signal using the ADC inputs to Vicon as described in the post above. This means we can apply a whole range of real recorded or simulated EMG signals, Sine waves or pulses. These can then be reviewed in Vicon, or in Labview by using the Export ASCII process.

Using Labview we can check the amplitude and frequency response of each sensor, and also check for delays between the sensors and the original signal, and between each sensor as well. We’re still ironing out a few issues with our technique, but hopefully we’re going to start using it soon before clinical sessions as a reassurance that the system is working as we expect it

[Elizabeth Parks]

Re: Validating wireless EMG results

We had a similar problem here in Birmingham. We the same set up as your lab, with a Vicon motion capture system and a Delsys Trigno EMG box.

We’ve recently completed a testing device that uses a Labview VI as a signal generator to send repeatable signals to 8 sensors at a time. We can compare this response to the sent signal using the ADC inputs to Vicon as described in the post above. This means we can apply a whole range of real recorded or simulated EMG signals, Sine waves or pulses. These can then be reviewed in Vicon, or in Labview by using the Export ASCII process.

Using Labview we can check the amplitude and frequency response of each sensor, and also check for delays between the sensors and the original signal, and between each sensor as well. We’re still ironing out a few issues with our technique, but hopefully we’re going to start using it soon before clinical sessions as a reassurance that the system is working as we expect it

This sounds like a great suggestion. How do you connect your Delsys base station to the LabView program. Do you use a sensorDAQ? Would it be possible to see your block diagram of the program? Thanks for your help.

[Philip Davenport]

Re: Validating wireless EMG results

The design came from the testing system for our previous EMG system, an MA300 – it was essentially a ‘Whisper’ EMG simulator with a set of contacts for the electrodes. The new design uses Labview to produce the signals, which are exported using a DAQ (NI 6219) to a single-sided PCB, with the sensor electrodes resting on separate square sections of the conducting surface, to measure up to eight sensors at once. There are a set of switches that can connect these contacts to look at CMRR and so on, but in general we apply the signal to one of the active electrodes only and record the signal using Vicon. We’ve found its best to use electrode gel to get a good surface contact between the electrodes and the PCB, and we’ve made foam spacers to locate the sensors onto the board.

The file reading VI is fairly simple, and generally just uses the express VIs to split up the individual waveforms from the ASCII file and look for the pulse applied or any other signal we put through, and then pull out the parameters that we want to look at (and should highlight any anything unexpected).

I’ve emailed you an outline of the VI, and a photo of the device with the PCB.