Scribd
Upload
100 views

Using Surface EMG

Surface electromyography (EMG) is used to measure muscle activation levels. It is difficult to identify muscles for each subject without experience. Experiments should include at least 100 practice trials to reduce variation between trials. The number of subjects, postures, and muscles studied should exceed 30-50. Statistical analysis of EMG data includes calculating averages, using t-tests to compare postures, and correlation tests for multiple variables. Sample experiments are provided to guide statistical analysis and interpretation of EMG experiments measuring muscle force, fatigue, and activation timing.

Uploaded by

Balachandran Kalidass
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
100 views

Using Surface EMG

Surface electromyography (EMG) is used to measure muscle activation levels. It is difficult to identify muscles for each subject without experience. Experiments should include at least 100 practice trials to reduce variation between trials. The number of subjects, postures, and muscles studied should exceed 30-50. Statistical analysis of EMG data includes calculating averages, using t-tests to compare postures, and correlation tests for multiple variables. Sample experiments are provided to guide statistical analysis and interpretation of EMG experiments measuring muscle force, fatigue, and activation timing.

Uploaded by

Balachandran Kalidass
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

Surface EMG

 Muscle Identification for each subject is a very difficult task and usually comes by
experience

 Before conducting an actual experiment, perform enough number of demos/trials (at


least 100). This will reduce time taken for the actual experiment, cost of having a subject
and performing the experiment without large deviation between trials

 No. of subjects x No. of postures x No. of muscles > 30~50

 The noise reduction/normalization/filtering details may not be specified in a journal (as


it is usually the experience of the person)

 Sampling rate can be found from any journal with similar experiment and chosen based
on amount of data required

 Surface preparation should be done as per protocol

Timer protocol for experiments

1. Ask person to start applying pressure and start recording

2. Ask person to apply maximum pressure at said time (E.g. At 3 seconds after start)

3. Ask person to hold the pressure as long as he can (the EMG response declines gradually)

4. Stop recording after the fatigue is reached where subject can no longer withstand (Sharp
decline in EMG response)

NOTE: The stop time can be different between the subjects as fatigue time is different.
The no. of values will be different for each subject. As we will be taking average of it, it
will not be a problem.
Choosing between Amplitude/Frequency/Timing values

1. Muscle force/Energy Expenditure – Use RMS, Maximum RMS value correspond to


maximum force.

2. Muscle fatigue – How long it takes for the muscle to fatigue. –Use Median Frequency -
Can be observed from the graph.

3. Activation timing – How long does the muscle take to respond – Use Amplitude-driven
event counter.

NOTE : For other applications, refer ‘Applications of EMG’ in following pages.

After experimentation – Statistical Analysis


P-Test
Consider 3 subjects (S), 4 postures (P) and 1 muscle for the experiment

Step 1: Take mean of all RMS values of S1 and P1. Similarly do for S1P2, S1P3.

Step 2: Similarly do for other subjects and constitute a table.

Posture 1 Posture 2 Posture 3 Posture 4


Subject 1
Subject 2
Subject 3
Average

Step 3: Take average for each posture.

Step 4: Now comparison can be done only between two postures at a time. So, do the p-
test (using formulae) for P1 and P2. The p-value obtained should be compared with the
table values and found if it there is significant difference.

Step 5: Similarly do the test between all combinations of Postures and plot the table

P1 P2 P3 P4
P1 x
P2 x
P3 x
P4 x
P1 P1 is neglected and other combinations are tabulated.

Step 6: Those combinations with higher significant difference should be marked bold or
highlighted.

Step 7: Inference can be given from the significant p-values. Two postures Px and Py may
be significantly different meaning there is some factor that is causing the difference
while transforming from Px to Py.

Step 8: When 2 or more values are significant, justification must be given to choose the
best posture. This is usually based on the application. For eg. For a triggering operation,
the highest peak of RMS value may be used to compare Px and Py. For a prolonged
operation, the larger fatigue stretch may be used.

Step 9: From p-test a hypothesis can be verified to be true or else the null hypothesis
remains valid (Read more on p-test).

Correlation test

When there are more variables, the calculation with p-test become more complex and it
is advisable to go with correlation test.

Similarly other statistical tests like regression test can be performed based on requirement.
Statistical calculations can be performed using excel.

Books 1) Thomas Quirk, Excel 2010 for Engineering, Springer (2014)

2) Conrad Carlberg, Statistical analysis Microsoft Excel 2010 (2011)

Sample experiment can be found in following pages.

Sample journals that used EMG

 D. Chakrabarti, IIT Guwahati - Usability is more valuable predictor than product


personality for product choice in human-product physical interaction (2014)
 V. Balasubramanian, IIT Madras - Detecting motorcycle rider local physical fatigue and
discomfort using surface electromyography and seat interface pressure (2014)
 G.G. Ray, IIT Bombay - Study on the variation of peak isometric strength and EMG
activity in static field-simulated lifting postures (2004)
Data processing procedure

Integrated data acquisition

After recording, select the required part of raw signal data

Click multiple add to data pad from menu and select sample size as per requirement (In this
case - select 1.0 second for every 1.0 minute)

In the data pad add variables like Mean, Median, RMS as required. Note the time interval (t=0
to t=1:00:12461 and check if it matches with selection)

From this data, statistical analysis is performed to get results and meaningful inferences are
concluded.
EMG response in Brachioradialis muscle during grip force in different postures

In some cases, the recording was stopped before actual fatigue, hence there was no decline seen.
After taking inferences, necessary reasoning should be given to select the best posture.

You might also like