A guide to peripheral physiology measurements using
the BrainAmp ExG MR – Part 1: Let’s focus on EMG
by Dr. Tracy Warbrick (Brain Products Application Specialist EEG-fMRI)
and Dr. Robert Stoermer (Brain Products Head of Technical Support)
Abstract
Measuring peripheral physiological signals during fMRI can add valuable information about physiological processes to your studies. Our BrainAmp ExG MR can be used to record bipolar signals such as surface EMG and ECG, and signals from peripheral physiology sensors measuring GSR, respiration, or acceleration can be recorded using auxiliary channels.
In a series of support tips, we will provide safety advice and methodological know-how that will help you to successfully record physiological signals in the MR environment with your BrainAmp ExG MR. The support tips will cover general considerations for ExG-fMRI measurements and then we will focus on specific measurements such as surface EMG, ECG, and sensors for peripheral physiology.
Introduction
Multimodal brain imaging studies using EEG and fMRI are very popular due to the advantage of simultaneous measurements using different temporal and spatial scales. Measuring physiological signals other than EEG during fMRI can also be valuable and can provide an insight into physiological processes that occur in parallel to changes in brain activity. For example, measures of arousal or muscle movements could provide complementary data. Our BrainAmp ExG MR can be used to record peripheral physiology signals alongside your EEG or as a standalone measurement. The BrainAmp ExG MR is a bipolar amplifier which means that the potential difference between a pair of electrodes is measured. This contrasts with the referential measurement principle used in our BrainAmp MR (plus) where the potential difference between each EEG electrode and a single reference electrode is measured. The 8-channel BrainAmp ExG MR has eight bipolar channels for signals such as surface EMG (sEMG), ECG, and EOG. The 16-channel BrainAmp ExG MR has eight bipolar channels and eight auxiliary channels. The auxiliary channels allow input from peripheral physiology sensors measuring GSR, respiration, or acceleration.
Recording high quality ExG data in the MR environment starts with good preparation, an optimal setup, and sufficient pilot testing. Therefore, in this series of support tips we will provide advice on how to safely setup ExG measurements in the scanner and share our methodological know-how to help you successfully record physiological signals in the MR environment with your BrainAmp ExG MR. In this first support tip we start with general considerations for ExG-fMRI measurements and then focus on sEMG-fMRI. In forthcoming articles, we will cover ECG-fMRI, using peripheral physiology sensors in the MR environment, and special applications using the BrainAmp ExG MR.
Overview
1. Safety
1.1. General safety considerations for ExG-fMRI measurements
1.2. Safety considerations for sEMG-fMRI
1.3. Studies requiring a customized setup
2. Optimising your sEMG-fMRI setup
2.1. Establishing optimised SNR for sEMG in MR-environments
2.2. Establishing the signal bandwidth for sEMG in MR environments
2.3. Optimising the gradient correction in BrainVision Analyzer 2
3. Our ‘top tips’ for successful EMG-fMRI
1. Safety
1.1. General safety considerations for ExG-fMRI measurements
All recommendations that we make throughout this support tip apply to measurements in MRI scanners up to 3 T and the safety guidelines and conditions for use should always be followed. You can find further information on our website. In the MR environment the BrainAmp ExG MR should only be used together with the ExG AUX Box (Figure 1) and special MR conditional electrodes and sensors (Figure 2).
For bipolar measurements (e.g. sEMG, ECG) Multitrode MR electrodes should be used because they have some special features that make them suitable for use in the MR environment: they have a current-limiting resistor, they are an incomplete ring to avoid induced eddy currents, and are bundled in a spiral tube so that the lead wire cannot come into direct contact with the participant (Figure 3). The Multitrode MR electrodes are intended for surface EMG and all recommendation that we provide in this article are for sEMG. For polygraphic signals we have three MR conditional auxiliary sensors: GSR MR Sensor, 3 D Acceleration Sensor MR, and Respiration Belt MR.
Figure 1: 16-channel BrainAmp ExG MR including ExG AUX box for connecting electrodes and sensors.
Figure 2 (left): The MR conditional sensors available for measuring peripheral physiology in the MR environment with the BrainAmp ExG MR. A. GSR MR, B. 3 D Acceleration Sensor MR, C. Respiration Belt MR, D. Multitrode MR (bound pair and single). Figure 3 (right): The Multitrode MR in detail, a bound pair of multitrodes is shown. A. incomplete ring to avoid induced eddy current, B. Current limiting resistor, C. Resistor value labels, D. MR conditional label, E. Plastic spiral tubing to prevent the lead wire having direct contact with the participant and to bind pairs of Multitrodes together.
The BrainAmp ExG MR should be placed at the foot end of the scanner bed for peripheral physiology measurements. By placing the BrainAmp ExG MR in the position shown in Figure 4, you can avoid routing electrode lead wires up the participant’s body and through the scanner bore towards the head and head coil. This minimises the safety risk that would be caused by running long cables through scanner bore and RF transmit field, and can also reduce negative effects on data quality, for example, cables routed over the trunk could pick up respiration movements and contaminate your signal of interest.
Figure 4: The BrainAmp ExG MR in the recommended position at the foot end of the scanner bed. In order to minimize electrode wire length, BrainAmp ExG MR, Powerpack and ExG AUX input box should be shifted head wards up to the edge of the scanner. The BrainAmp ExG MR, Powerpack and ExG AUX input box can either be placed in between the knees on a knee cushion or an MR safe table that is placed over the legs of the participant. The measurement in the example is sEMG from the lower arm using custom length Multitrode MR lead wires.
1.2. Safety considerations for sEMG-fMRI
Surface EMG signals can be measured at a wide variety of sites on the body and in contrast to EEG measurements, where the electrode position and lead wire routing is determined by the electrode cap, the position and routing of sEMG electrodes is up to the experimenter. This results in a great deal of variability in possible configurations. As such, sEMG-fMRI measurements can be considered riskier than EEG for the following reasons:
These factors increase the possibility of heating in the electrodes and lead wires; this is something that we want to avoid so we can protect the participant and the BrainAmp ExG MR from harm. Now, our aim is not to scare you and deter you from doing EMG-fMRI measurements! sEMG-fMRI measurements can of course be done safely, and in this article we will explain how. The above list of risks can be reduced by following some simple recommendations:
1.3. Studies requiring a customized setup
The standard Multitrode MR has a 40 cm lead wire. This length is sufficient for lower limb sEMG (Figure 4); however, you might want to measure from other sites on the body. If we consider the recommended position of the amplifier at the foot end of the patient table, the variety of possible positions for the sEMG-fMRI electrodes, and the need to route the cables in a straight line, it becomes clear that custom cable lengths may be required for each individual sEMG setup. This also requires the height of the test population to be taken into consideration (e.g. children versus adults). We are happy to provide custom length MR conditional electrodes for sEMG. However, in contrast to standard EEG equipment, customised solutions are subject to a liability waiver because the experimenter is solely responsible for the site-specific implementation. Please get in touch with our Technical Support team if you have any questions about appropriate electrode lead wire lengths for your study.
2. Optimising your sEMG-fMRI setup
In addition to the safety aspects, it’s also necessary to keep in mind that bipolar measurements in the scanner can be challenging with respect to signal quality. But don’t worry, we will explain the secrets of successful sEMG in the scanner!
While EEG recordings in the MR environment work easily out of the box, successful sEMG in the scanner is based on three pre-conditions:
1. Optimised signal to noise ratio (SNR) in the raw signal going into the BrainAmp ExG MR
2. Bandwidth optimisation to facilitate using the BrainVision Analyzer 2 AAS gradient artifact correction algorithm
3. Optimal correction strategy using the Analyzer 2 AAS gradient artifact correction algorithm
We will take each of these pre-conditions in turn and explain how to optimise them.
2.1. Establishing optimised SNR for sEMG in MR-environments
When considering SNR optimisation, we assume that a crosstalk free sEMG is the signal of interest and the noise is caused by the scanner gradient system.
The first point we should consider is impedance at the skin-electrode interface. Due to the higher frequency band of interest, skin electrode impedances for sEMG are more important than for scalp EEG. We advise that impedances are approximately 5 kOhm for sEMG (Fridlund & Caccioppo, 1986). Preparation of the measurement site is also important and there are some considerations specific to sEMG. For example, limb sEMG is often carried out in regions with a thicker and drier epidermis than scalp EEG, so careful preparation using alcohol and high chloride abrasive gel is important and we recommend using Abralyt HiCl. Achieving the recommended impedance level is facilitated by the BrainAmp ExG MR impedance measurement circuit which gives precise impedance values for the positive (+) and negative (–) lead wire of each sEMG channel separately. Separate measurement of the + and – channels is important because they must be balanced to facilitate BrainAmp ExG MR common mode suppression in the best possible way.
In addition, the impedance of the ground (GND) electrode must be as low as possible and the lead wire paths of the + and – wires must be as close to each other as possible. Our pairs of Multitrode MR electrodes are bound together with a spiral tubing to make sure they stay close together.
The position of the recording electrodes is critical to optimising the SNR in the MR environment because the inter electrode distance affects not only the sEMG amplitude and possible crosstalk, but also the gradient artifact amplitude, thus signal AND noise. Consequently, when measuring sEMG in the scanner there is trade-off between maximised EMG amplitude and minimised gradient artifact amplitude. The further apart the electrodes are, the larger the gradient artifact, so a classical placement that works in the lab doesn’t generally work in the scanner, the electrodes need to be closer together. While EEG electrode positioning is given by the cap layout, sEMG electrode positioning has more degrees of freedom and therefore a greater impact on the overall signal quality. A good place to start for meaningful electrode positions and electrode distance is the SENIAM project: http://www.seniam.org. Once you have a good starting point, the fine tuning must be done in pilot sessions until the gradient correction results and EMG signal properties match the requirements of the study.
2.2. Establishing the signal bandwidth for sEMG in MR environments
While most EEG research is done in a frequency range below 60 Hz, for sEMG we required a higher signal bandwidth after gradient artifact correction. To achieve this, optimal preconditions for artifact removal using the average artifact subtraction (AAS) (Allen et al., 2000) method must be established. The AAS method uses a template subtraction approach and to make sure that the average artifact template closely matches the artifact to be subtracted, precise timing and recording of the artifact is necessary. The AAS approach is explained in Figure 5 and the preconditions are described below.
- 5 kHz sampling rate
- 0.5 µV/bit amplitude resolution for the analogue to digital converter (ADC)
- AC-coupled acquisition with a low cut-off hardware filter with a time constant of 10 s
- 250 Hz hardware high cut off filter. The 1 kHz low pass filter would be desirable from an EMG perspective, however, when recording in the MR environment one must consider the gradient fields and on a whole-body scanner the gradient fields would often saturate the ADC. The 250 Hz high cut off is therefore recommended.
- Make sure that the correct value for the resistors in your electrodes are entered in the workspace. This value will be subtracted during the impedance measurement to give accurate impedance values. The correct resistor value can be found on a small, white label close to the electrode and the connector on the Multitrode MR (Figure 3).
- Volume or slice markers are sent from the gradient system of the scanner and can be recorded with the BrainAmp ExG MR using the trigger input on the USB 2 Adapter (see picture on the right). The TriggerBox can also be used with the BrainAmp MR system to receive triggers from the gradient system. The TriggerBox supports the conversion of optical triggers from the gradient system into TTL pulses for the USB 2 Adapter. It also helps to handle and merge triggers arriving from different sources. This can be very useful if you have stimulus and response triggers in addition to triggers from the scanner.
- When possible, slice-wise correction of the gradient artifact is preferred (Figure 5). This requires that slices are of an equal length, and that the duration of the slices is an integer multiple of 200 µs (the duration of one sampling point with the recommended sampling rate of 5 kHz). If either of these conditions are not met volume-based correction is recommended.
- However, it is worth keeping in mind that volume-based correction also benefits from considering the temporal aspects of the fMRI sequence; the TR (repetition time) and slice duration should be an integer multiple of 200 µs when possible (Mullinger et al. 2008).
Figure 5: The sliding average concept used in average artifact subtraction (AAS) for gradient artifact correction. In part A, volume 4 is to be corrected by making an artifact template from volumes 1-7. In part B, volume 5 is to be corrected by making an artifact template from volumes 2-8. In part C, volume 6 is to be corrected by making an artifact template from volumes 3-9. This sliding average strategy is continued throughout the data, so the template for each volume is created from neighbouring volumes. A total of 21 volumes is recommended – the volume to be corrected plus 10 volumes before and 10 volume after. When the imaging volume is used to form the template the duration of the data included is 21 times the TR (repetition time). For example, with a TR of 2 s the data used in the template would span 42 s. If slices are used this overall duration is much shorter. For example, a slice duration is usually less than 100 ms (depending on the sequence), so the data used in the artifact template for 21 slices of 100 ms would be 2100 ms. This shorter duration means that the template is less susceptible to fluctuations in the gradient artifact over time, for example from movement.
2.3. Once the setup matches requirements, the gradient correction in BrainVision Analyzer 2 needs to be optimised
A comprehensive demonstration of how to use the MR Correction transformation in Analyzer 2 is provided in our webinar on handling scanner related artifacts, detailed instructions can also be found in the Analyzer 2 user manual. Below we will give an overview of points to consider when processing your sEMG data.
3. Our ‘top tips’ for successful EMG-fMRI
Conclusion
We have covered the most important points to help you perform safe sEMG-fMRI measurements and obtain high quality sEMG data in the scanner. We hope you found this support tip helpful and if you have any further questions on this topic, please feel free to contact our Technical Support team. Remember that this is just the first in a series of support tips on ExG-fMRI measurements so stay tuned.
References
Allen, P. J., et al. (2000).
A method for removing imaging artifact from continuous EEG recorded during functional MRI.
Neuroimage 12(2): 230-239.Fridlund, A. J. and J. T. Cacioppo (1986).
Guidelines for human electromyographic research.
Psychophysiology 23(5): 567-589.Mandelkow, H., et al. (2006).
Synchronization facilitates removal of MRI artefacts from concurrent EEG recordings and increases usable bandwidth.
Neuroimage 32(3): 1120-1126.Mullinger, K.J. et al. (2008)
Improved artifact correction for combined electroencephalography/functional MRI by means of synchronization and use of vectorcardiogram recordings.
J. Magn. Reson. Imaging, Vol. 27, Issue 3: 607-616.