I am currently working on artifact removal from ambulatory EEG for seizure detection of an epilepsy patient. I have gone through some literature related to this topic but found almost all of it to use some kind of motion sensors; e.g. accelerometer, gyroscope, etc. to track the movement/motion artifacts and hence use suitable techniques to remove them. However, I am wondering if I can do this without using the sensors, since not every ambulatory EEG recorder would come with such sensors. I would like to work fully from the signal processing point of view in digital domain and I don't want to interfere/modify the recording setups. Hope I made it simple to understand. Thanks in advance.
A way to significantly reduce (actually remove) motion artifacts in ambulatory EEG is to use amplifiers with active shielding technology. We use Mobita amplifiers from TMS-international (http://www.tmsi.com/)
We recently reported on our experiences in this paper:
Mobile EEG in Epilepsy.
Jessica Askamp, Michel J A M van Putten
International journal of psychophysiology: official journal of the International Organization of Psychophysiology (Impact Factor: 3.05). 09/2013; DOI:10.1016/j.ijpsycho.2013.09.002
there are 2 types of motion artefacts. the first one is connected with motion of the patient, and the second one - with the scalp muscles. The first type could be studied using the video recordings which are done together with EEG - it is not easy task, but doable. The second type is connected with specific frequency range and one may find it as a "constant" trace during the recording period in the frequency-time domain (it looks like the el. current trace around 50 or 60 Hz). One may see it clear applying for example Gabor filtering space in time with a sequence of central frequencies covering the working frequency interval - let say from 1 until 128 Hz. For the precise frequency band one need to consult with neurophysiologist.
Artefacts have physiological origins (muscle activity, heart, EOG ...) or physical origins (movement of the electrodes on the scalp, electrical noise like 50Hz,....). Lot of studies are using ICA components analysis to rejects artefacts mainly in the motor control domain or coupled EEG-fMRI studies. In some cases classical bandpass filtrering could also work.
Except all of the above technologies, your can use adaptive filter with a reference signal correlated with the motion artifact. Meanwhile, one of the reference signal we used is the electrical tissue impedance(ETI). Maybe you can use 3-axis accelerometer as well.
I have developed an open source matlab toolbox for the automatic EEG artifact rejection, as published in: http://www.mdpi.com/1099-4300/16/12/6553
If you are interested, please visit: http://nadiamam.wix.com/eawica
- Badges
- Science topic
- Similar topics
- Journal Impact Factor