Can I collect/use EEG data while running an experiment in a KINARM Lab?

A number of BKIN customers have successfully recorded EEG data from subjects while using a KINARM Lab (Torrecillos et. al, 2014). Depending on the goals/needs, there are a couple of considerations about what is possible and how to achieve it. Post-experiment analysis: Because the KINARM Lab has fewer data acquisition channels (i.e. analog inputs) than is typically required for EEG recordings, a separate EEG recording system is needed. However, that can result in separate data files: one with EEG data and one with KINARM data. Because no two clocks are every perfectly synchronized, this separation of data requires post-experiment synchronization for proper data analysis. There are a number of possible approaches that can be used to solve/avoid this problem.

1. The most common approach is to use digital I/O channels to sync the EEG data with the KINARM data. KINARM Labs include a number of digital output channels, so this approach requires an EEG system that can record external digital inputs. The KINARM Lab can be programmed to use these channels in multiple ways, for example:
   • To send a pulse on digital channel 1 to indicate trial start and to also send a pulse every xxx ms on digital channel 2 to provide a “clock” signal
   • To send digital channel 1 high for as long as EEG recording is desired and to also send a pulse every xxx ms on digital channel 2 to provide a “clock” signal
   • Any other combination of digital channel controls (i.e. the KINARM Lab can be programmed to control digital pulses at a max rate of 4 kHz).
2. It is also possible to use digital I/O channels to send digital words (e.g. 8-bit or 16-bit signals). As with the prior solution, this approach requires an EEG system that can record digital signals that 8 or 16-bit wide.
3. Another option is to connect the EEG and KINARM Labs using Ethernet and send data or synchronization pulses via UDP. This approach requires an EEG system that can either stream out data over UDP, or receive and record data over UDP. The KINARM Lab can be programmed to either send and/or receive customized UDP packets, which can be used to:
   • Send synchronization packets, analogous to the digital I/O described above.
   • Send all KINARM data from the KINARM Lab to the EEG so that the EEG system saves all data in a single data file (thus avoiding the need to synchronize data post-experiment).
   • Send some pre-processed EEG data to the KINARM Lab so that the KINARM lab saves all data in a single data file (also avoiding the need to synchronize data post-experiment)..

Online control of a KINARM robot using EEG: It is possible to control the KINARM robot, in real-time, using EEG data. As per above, there will need to be an external EEG system capable of recording and analyzing the EEG data. There are then a couple of options for sending a “command” from that EEG data back to the KINARM Lab to provide real-time control.

1. Send the command from the EEG to the KINARM Lab via analog signals. This approach requires an EEG system with analog outputs. KINARM Labs are available with an optional analog input module, so the EEG system would have to analyze the data and then convert that analysis into a desired “command” signal (e.g. force in the x and y directions) as analog outputs. The KINARM Lab would then have to be programmed to use those analog inputs as input commands.
2. Send the command from EEG to the KINARM Lab via UDP. This solution is similar to the analog signals solution, but the communication would be via Ethernet and UDP. As per above, the KINARM Lab can be programmed to receive and interpret and UDP packet, so this approach would require an EEG system that is capable of turning the desired “command” into a UDP packet.

The following are some comments from another KINARM customer who has successfully recorded EEG with a KINARM Exoskeleton:   Here are the common artifacts in my EEG-KINARM data:

• Artifact caused by KINARM: 50/60 Hz ambient electrical noise -
  • since the participant's head is very close to the subject display, very obvious electricity artifacts can be observed. However, this can be removed by applying the notch filter in these frequency lines.
    • < IEB comment - the magnitude of any 50/60 Hz electrical noise from the display will obviously be dependent upon the display technology. This comment comes from a KINARM Lab that has an older, non-LED subject display. Newer KINARM Labs use LED displays, which are likely to have far less 50/60 Hz electrical noise. >
• Artifacts caused by movement: sweat artifact & electrode movement artifact -
  • The movement tasks sometimes make participant sweat... and then there will be a decreased amplitude in very low frequency (0.25 - 0.5 Hz ). Therefore, you have to always try to make the airflow around the KINARM better. Especially the participants are squeezing in such a small space when they do the task.
    • < IEB comment - in 2016 BKIN Technologies introduced a KINARM Exoskeleton in which the motors are below the shoulder, rather than beside head. This change opens up the space around a subject's head, which should improve airflow >
  • Electrode movement is also a difficult part, since most KINARM tasks are gross movements... This artifact can be random or rhythmic, depends on the tasks. In my experience, we cannot completely avoid it, but the artifacts can be largely reduced if you:
    1. make sure the cap really fit the participants head... the contact surface must be very flat.
    2. ask them to shake/nod their head a bit, and move their arms before the testing. In the meanwhile, you observe how the artifact looks like, and try to fix the cable/cap in a better way.

April 2020 Publication Using Kinarm and EEG The Motor Neuroscience lab at Michigan State University just published a paper using the EEG in conjunction with the Kinarm Lab. In this study, "Participants performed a bimanual center-out reaching task in which a visuomotor rotation was applied to the right hand while the left hand did not receive visual feedback of its movements. ... Electroencephalography (EEG) recordings during the task showed that spectral power in the high and low beta frequency bands was elevated early in exposure, but decreased throughout learning" (Desrochers et al., 2020). You can find the article here, or in our publications database.