After investing months of time, funds, and personal effort into your electrodermal activity (EDA) study, you begin combing through your data recordings when something catches your eye. Several of the recorded signals don’t look quite right. Rather than a narrow baseline in the signal graph, the baseline is thick and fuzzy, a sure sign of signal noise. To make matters worse, the recordings are broken by random peaks, artifacts unrelated to the EDA signal you sought to record. Noise and artifacts can corrupt research data, often rendering it useless. While it may be possible in some cases to clean up and salvage recorded data, such results often mean going back to square one and starting over.
The best option is to avoid bad data from the beginning. Several best practices can save researchers untold effort and expense by avoiding the pitfalls that can result in noise and artifacts. Noise and artifacts can be defined as unwanted changes to a signal caused by sources other than the target signal during recording. Noise and artifacts can originate from a range of sources—environmental (interference from other electromagnetic sources) or biological (signal changes from muscle activity in a participant), to name a few. The key is not to wait until all your data is recorded to tackle signal problems. Careful planning and testing procedures when designing and experimenting can avoid later headaches.
Most artifacts are caused by some form of electrical impedance interfering with a clean connection between the participant and the electrodes linking the data-gathering equipment. Sources could include dead skin or hair at the connection sight, dirty or improperly prepared electrodes, or electrodes that have come loose due to movement.
Using an impedance detector can help identify problematic signals before they corrupt data. An example of this can be found in a 2020 study at Ghent University in Belgium that examined how expert and novice map readers managed their cognitive loads using electroencephalogram (EEG) and eye tracking data. Researchers used a BIOPAC EL-CHECK impedance detector to ensure a good skin-electrode interface before recording data with AcqKnowledge software. This allowed the research team to keep the whole EEG circuit (ground, active EEG, and reference electrodes) at less than the recommended 10 K ohms (Herman, 2015; Teplan, 2002) to prevent unwanted signal noise.
Strategies for avoiding artifacts can vary depending on the type of signal being recorded. Some signals, such as EEG, may require light abrading of the skin at the point of electrode attachment for optimal conductance, while other signals, like EDA, do not require skin abrasion. Proper site preparation is recommended for most signals, skin should be clean and free of obstructions, such as excess hair. When gel is to be applied, make sure it is the correct type for the electrodes being used. Finally, the preparation of study participants can be critical to successfully recording clean data. In general, participants should be in an environment that is free of distractions and stimuli originating outside the study (sudden noises, distracting conversations, etc.).
For detailed information on how to proactively prevent signal noise and artifacts, check our AcqKnowledge Bootcamp webinar in which we discuss how to set up experiments to ensure quality data gathering. You can also consult our forthcoming guide with detailed instructions for the collection of optimal data. Don’t wait until it’s too late; eliminating signal noise should be the first step in setting up a successful study.
Do you need help eliminating artifacts and signal noise in your recorded research data? BIOPAC’s regional sales representatives are ready to assist with tools and strategies to ensure you get only the best data possible. Contact us today!