by Dr. David Schubring
Scientific Consultant (Brain Products)
Joining different labs to run the same study protocol has many advantages, e.g., including more participants from more diverse demographic and geographic backgrounds, comparing results between labs, benefiting from networking and knowledge transfer between colleagues, and ultimately a better generalizability of the results. On the other hand, multicenter studies also pose several challenges.
In this article we focus on one of these challenges: Aligning hardware requirements in the different labs. It is important to stress that the gold standard is to use identical hardware with identical settings in all labs.
Using different hardware across labs: Impacts and how to minimize them
In short …
The recorded signal from all Brain Products EEG amplifier families (BrainAmps, actiCHamps, LiveAmps) are interchangeable and only differ as technically unavoidable. As a general comparison, the stationary amplifiers actiCHamp (Plus) and BrainAmp offer the widest array of possible configurations, while the LiveAmp is optimized for versatility and mobility. However, all three amplifiers still offer research grade signal quality.
|BrainAmp DC||actiCHamp (Plus)||LiveAmp|
|Common mode rejection ratio at 50/60 Hz||≥ 110 dB||≥ 100 dB||> 80 dB|
|Input noise for DC||≤ 1 μVpp||≤ 2 μVpp||≤ 2 μVpp|
Similarly, the signals from all our electrodes will be largely comparable, but the different electrode concepts also offer very specific profiles and advantages, e.g. in terms of noise immunity (active electrodes) or preparation time (dry electrodes). Basically, we distinguish three types of electrodes:
|Gel based active||as in actiCAP snap/slim and its predecessor actiCAP|
|Gel or sponge-based passive||as in BrainCaps and R-Nets|
|Dry electrodes||as in CGX Quick Systems and actiCAP Xpress Twist|
The more heterogenous the electrode types are the greater the variance in noise immunity. The main caveat is that we generally not recommend mixing dry electrodes with the other two electrode types in the same multicenter study, as the construction principles differ too much.
The minimum technical requirements and settings must be defined before the study. As many of these settings as possible should be standardized across labs during recording. If not possible during recording, some settings can also be adjusted offline. Finally, one should also be aware which differences in settings cannot be avoided.
1. Frequency bandwidth, filters, and sampling rate
The required bandwidth depends on the kind of signal (EEG, EMG), focus of the study and specific guidelines (e.g., Keil et al., 2014; Fridlund & Cacioppo, 1986). As a rule of thumb, the sampling rate should be at least five to ten times higher than the frequency of the signal of interest for best results (Weiergräber, Papazoglou, Broich & Müller, 2016). Theoretically, frequencies of up to half of the sampling rate can be analyzed (Nyquist frequency), but Brain Products amplifiers also employ additional filters at roughly 1/4th of the respective sampling rate to avoid aliasing artifacts. To get a linear frequency response, the filter cutoff should be spaced sufficiently apart from the frequency of interest. E.g., for classical ERP studies analyzing signals of up to 50 Hz, a minimum 250 Hz sampling rate is recommended. If additional signals such as high gamma (up to ~80 Hz) or electromyography (up to ~500 Hz) should be included, the sampling rate must correspondingly be higher.
The maximum available sampling rate differs between amplifiers: The LiveAmp supports a sampling rate of up to 1 kHz (with 32 channels), BrainAmps up to 5 kHz, and actiCHamp (Plus) up to 100 kHz (with 32 channels). This sampling rate should be equalized across labs, either during recording in the BrainVision Recorder workspace or after recording by means of downsampling.
In addition to the sampling rate, hardware (only for the BrainAmp DC and BrainAmp MR plus) and software filters can be set during recording. In case of the two hardware filter options of the BrainAmp DC and BrainAmp MR Plus (250 Hz and 1000 Hz), these should also be spaced sufficiently apart from the signal of interest and be set at least 1/4th of the sampling rate1.
No additional raw data saving filters during acquisition are recommended, as these limit the choices of filtering after acquisition.
2. Measurement range | resolution | bit depth
2.1. BrainAmp amplifiers
The resolution can be changed in the BrainAmp DC and BrainAmp MR plus. This resolution also affects the measurement range: 0.1 µV or 0.5 µV per bit, resulting in a measurement range of ±3.28 mV or ±16.384 mV with a 16-bit A/D converter2.
If leaving the measurement range is not an issue, we generally recommend using the highest possible resolution (0.1 µV and ±3.28 mV range) in the BrainAmp.
2.2. actiCHamp (Plus) amplifiers
These amplifiers have a fixed resolution of ≈0.0487 µV per bit resulting in a measurement range of ±409.6 mV with a 24-bit A/D converter.
2.3. LiveAmp amplifiers
These amplifiers have a fixed resolution of ≈40.7 nV / bit and a measurement range of ±341.6 mV with a 24-bit A/D converter. All in all, the signals will still look similar even when recorded with different resolutions, however, one should still be aware of the differences.
3. Electrode impedance & amplifier input impedance
All labs should work with the same recommendations for desired impedance levels of electrodes. When using actiCAP slim/snap (or its predecessor) with actiCHamp (Plus) the default thresholds are:
|good impedance||< 25 kOhm (green LED)|
|acceptable impedance||25 – 50 kOhm (yellow LED)|
|bad impedance||> 50 kOhm (red LED)|
However, specific recommendations about impedance thresholds depend on many things: electrode type (active vs. passive), amplifier input impedance, desirable preparation time, environmental noise, signal to noise ratio, and so on.
Rather than using individual thresholds in each lab depending on its specific setup, we suggest using one common impedance threshold all labs should work with. This also balances out participant fatigue, e.g., if one lab needs 60 minutes to prepare a subject and another lab needs 10 minutes, fatigue effects might overshadow the setup differences. It is also a good idea to pilot the study in the different labs to see if the preparation time and resulting impedance is similar.
4. EEG montage, including reference and ground
Most of our electrode caps follow the international 10/20 system, i.e., the relative electrode positions should be the same in the different caps.
Moreover, the BrainAmp and LiveAmp have an additional fixed reference electrode, while the actiCHamp (Plus) can choose a reference freely from one of the available EEG sensors. That means that the actiCHamp (Plus) has one sensor fewer than the other two. For mixed setups, we recommend re-referencing the data after recording to the same reference montage and removing sensors that are not shared across labs. In this process the reference sensor can also be “regained” as the negated new reference.
Tip: Head over to “Choosing your reference – and why it matters” for more information on re-referencing.
5. BrainVision Recorder version & Recorder workspace
It is always a good idea to use the newest BrainVision Recorder version to get all the latest bugfixes and updates. Multiple labs should also use the same Recorder workspace if they use the same amplifier type, but for different amplifiers, individual Recorder workspaces are needed.
6. The ‘Rest of the setup’
There are also other factors involved than the Brain Products hardware, including stimulation software and environmental noise. While we cannot directly comment on that, it might be a good idea to check these, too.
Specifically, the latency of the stimulation setup might need to be tested to avoid one lab having e.g., 10ms trigger latency and another 50ms, leading to systematic differences. One possibility to verify that latency is with the StimTrak. It is also a good idea to pilot the study in the different labs to see if any other unforeseen differences occur and to equally train the research staff (e.g., one lab having much more environmental noise than the other, which one might be able to optimize before the study starts). The differences between setups also has different impacts depending on the paradigm, see e.g. Melnik et al. (2017) finding bigger setup differences for steady state visually evoked potentials (SSVEP) than for other paradigms (possibly also because of stricter trigger latency requirements for SSVEP).
Another thing to check in advance would be which information to include in the study protocol and method section ahead of time, to see if it is aligned before conducting the study. One such guideline would be Keil et al. (2013).
In summary …
We recommend recording with equal settings if possible (BrainVision Recorder version and workspace, EEG montage, electrode impedance) and with the least restricted settings if not possible (sampling rate, filter, resolution), to better align these settings later during preprocessing (resampling, digital filtering, re-referencing).
However, this is only a short overview of potential technical issues and solutions. If you have specific questions regarding your multicenter study, feel free to contact the Brain Products Technical Support team at any time.
1 If the hardware filter is insufficient for the sampling rate, an additional digital filter is employed prior to downsampling in the Recorder software.
2 It also offers an option of 10 µV resolution and ±327.68 mV measurement range, but that is intended only for technical measurements and not for EEG.
Fridlund, A. J. & Cacioppo, J. T. (1986).
Guidelines for human electromyographic research.
Psychophysiology, 23(5), 567-589. https://doi.org/10.1111/j.1469-8986.1986.tb00676.x
Keil, A., Debener, S., Gratton, G., Junghöfer, M., Kappenman, E. S., Luck, S. J., … & Yee, C. M. (2014).
Committee report: publication guidelines and recommendations for studies using electroencephalography and magnetoencephalography.
Psychophysiology, 51(1), 1-21. https://doi.org/10.1111/psyp.12147
Melnik, A., Legkov, P., Izdebski, K., Kärcher, S. M., Hairston, W. D., Ferris, D. P., & König, P. (2017).
Systems, subjects, sessions: to what extent do these factors influence EEG data?
Frontiers in human neuroscience, 11, 150. https://doi.org/10.3389/fnhum.2017.00150
Weiergraeber, M., Papazoglou, A., Broich, K., & Mueller, R. (2016).
Sampling rate, signal bandwidth and related pitfalls in EEG analysis.
Journal of neuroscience methods, 268, 53-55. https://doi.org/10.1016/j.jneumeth.2016.05.010