MONTREAL NEUROLOGICAL INSTITUTE		MEMBERS
McGILL UNIVERSITY		PUBLICATIONS

EEG-fMRI intro page

Methods of analysis: combining several HRFs

HRF projection

The MRI scanner is a difficult environment to attain good quality EEG. In our experiments, we depend on the quality of the EEG signal to be able to visually identify epileptic spikes in our patients, which are used as events in the analysis. Another situation where EEG in the MRI scanner is evaluated is for the assessment of event related potentials (ERPs), identified as small changes in the EEG in response to stimulus. Careful attention must be taken during EEG equipment setup to ensure that artifacts are minimized. Even after considerable experience and practice, a perfect EEG signal cannot always be guaranteed.

One issue that is important in all fMRI studies, but especially EEG-fMRI studies, is subject’s cooperation. Within the scanner the EEG is especially sensitive to artifact. Muscle artifacts must be minimized and excessive movement artifacts in the EEG can be obstacles for proper removal of the artifact caused by the RF pulse from the scanner.

There are three major artifacts resulting from recording the EEG in the scanner: 1) the ballistocardiogram artifact, 2) artefacts caused by movement of the electrode wires inside the magnetic field (even if movements are very small, artefacts are large as a result of the high field), 3) a high amplitude artifact caused by the RF pulse during scanning. Movement artifact and the ballistocardiogram artifact can be minimized by the EEG setup, and the RF artifact can be filtered out with appropriate software. There are also other artefacts that may be caused by the magnetic properties of some materials, such as the electrode paste.

EEG Setup

Many factors can help improving the quality of the EEG, from the type of conductive paste applied for placement of electrodes, to the way the electrodes and wires are connected to the amplifier.

We use individual electrodes, with three-meter long wires grouped in cables, connected to an in-house electrode box. After electrode placement, all wires are pulled together at around CPz-Pz, from where they are grouped in bundles of six in one of the cable sheaths. It is very important that the wires run as straight as possible over the scalp from the electrode to the cable entry, to avoid any loops. The length of each electrode wire must be therefore adjusted on an individual basis. In addition, electrode impedance should not be greater than 5 KΩ. Special attention should be given to reference and ground electrodes.

We use cotton bandages to wrap the patient's head, adding extra padding to the occipital areas. Because of the length of our experiment, it is important that the patient is as comfortable as possible while lying on the electrodes.

When the patient is placed on the scanner bed, we have to ensure comfort, keep the wires still and minimize head movements. Significant discomfort is caused by the pressure points at the electrodes. This can be in large part alleviated by the use of a plastic bag filled with very small polystyrene spheres, in which a vacuum is obtained by air suction. A bite bar is not recommended, as it may cause discomfort and muscle artifacts over the temporal electrodes.

Opinions vary on whether it is acceptable to have the amplifier within the bore of the magnet (some amplifiers have longer and shorter connectors). In our setup, the amplifier is placed on a table about one meter behind the back end of the bore.

In order to minimize movement of the electrode wires before they reach the amplifier, we place them on a wood board and we cover them with sand bags. We also coil our electrodes wires (the sheaths are twisted around each other until the form a fairly tight, straight, and even coil).

Removing Artifacts

As previously mentioned, there are three main artifacts in the EEG:

Gradient artifacts: are caused by the currents induced in the electrodes and leads by the rapidly changing magnetic fields. Continuous EEG acquisition during EEG-fMRI experiments requires an amplifier with a sufficiently large dynamic range, such that there is no saturation. We use an amplifier with a sampling rate of 5000Hz.

Ballistocardiogram artifact: This consists of deviations following each heartbeat and possibly originates from small movements of the head or the electrodes following each pulsation because of fast movement of the blood in the arteries. It has been noted from the first reports as one of the main problems when recording EEG in an MR scanner and can be removed by averaging and subtraction, adaptive filtering, wavelet filtering or Independent Component Analysis. It is significantly worse in a scanner with higher field strength.

Another important issue can be the movement artifacts. Some filtering algorithms my have difficulty distinguishing between gradient pulse artifact and movement artifacts, making the process more troublesome.

There are different artifact removal methods and for a review on the available techniques, please refer to Bénar et al (2003) and Gotman et al, 2005.

 

Identifying the Spikes

After filtering the EEG, a neurophysiologist who is experienced in interpreting EEGs recorded inside the scanner reviews the tracing and visually identifies each epileptiform discharge, discriminating them according to the field of electrodes involved and the morphology of the event. These events will be then used for creation of the statistical maps.
 

HOME