Artifacts & Artifacting:

Regardless of whether you are doing a MiniQ or a full qEEG you should be able to recognize artifact so that you can either edit it out, if you have editing abilities, or repeat the collection run in the case of the MiniQ. The MiniQ is capable of absorbing some modest artifact and contains indicator lights in the report if the artifact is too high. The examples below are from a full qEEG with simultaneous acquisition but they apply to recording two or four channels sequentially for MiniQ as well. The details on editing are not necessary to know for those doing MiniQ and can be skipped over.

Whenever an individual moves during the recording of an EEG it generates distortions in the EEG signal, and these become amplified. These distortions are known as artifact. Blinking, swallowing and clenching the teeth are famous for generating artifact. Poor electrode connections also produce artifact. When poorly connected electrodes move on the head from subtle body motion it can generate a spike or "pop" in the EEG record. All of the above inspire qEEG collectors to strive to have individuals sit as still as possible during the recording of an EEG to reduce artifact.

The artifact generation can be diabolically subtle at times. Eye flutter can generate frontal artifact that often looks like delta waves or even spike and wave patterns. According to Neidermeyer, eye movement and flutter can generate false EEG waves up to frequencies as high as 13HZ and even higher. Subtle rhythmic movement of the head generally generate delta like waves. Some artifacts can be so subtle at times that it is necessary to refer to experts to identify them. My expert is Marvin Sams, who invented the Electro-cap that so many of us use to collect EEG. He has helped me identify some of the artifact represented below and I keep them in a file named "Marvin's Hall of Fame. "

Eye blinks tend to generate theta frequency waves, while eye movement tends to generate delta type wave as seen in the two records below.

Eye Blink

At 1:21 notice the large deflection in the EEG in the frontal leads Fp1 and Fp2. You can see it also migrates into F3 and F4.

Eye Movement Subtle

You can see a gentler and rolling deflection in the frontal leads in this example below.

Eye movement that is up and down produces frontal excursions in the EEG record that goes up and down.

Up and Down Eye Movement Extreme

Eye movements laterally, left and right, generate waves that move together and away from each other.

Lateral Eye: Note Crossover at F7-F8

Muscle tension looks spikey and continuous. It stands out against the background of normal EEG. Fp1, Fp2, T3 and T4 stand out in this record, although more subtle muscle tension can be seen at F7 and F8.

As mentioned previously, eye flutter produces delta type waves that are hard to distinguish from real EEG (Fig).

Key Artifact locations

Eye movement tends to appear at Fp1-Fp2, F7-F8, and F3-F4 depending on what locations are present in your map. In power or magnitude maps it tends to dominate the delta frequencies. Muscle tension tends to be prominent in the same locations but shows up in the beta frequencies and often at T3 or T4, or both, as well. Eye blinks tend to show up most in the theta frequencies. Below, the magnitude, coherence, phase and symmetry values indicate the probability that the delta has artifact in at least the frontal and temporal locations and the artifact light has been triggered. This indicates that the raw data collected should be played back in real time for review to determine the significance of any artifact that might be present. The same procedure can be done in the editing screen with EDF files that have been collected for the analysis.

Poorly connected electrodes can generate EEG that looks very high amplitude and unstable with large surges of power as seen in the graph below. This shows record shows an electrode that is partially loose.

They can also produce excess beta and excess theta. The excess beta that appears in the case of loose electrodes is frequently a result of 60HZ artifact that also produces the humming sound that comes from fluorescent lights, and can migrate down into the beta ranges. Both 60Hz hum and excessive muscle tension can migrate into the lobeta range and generate false magnitude or power readings.

Muscle tension can be very frustrating to deal with and there is no way around it sometimes. Many people have tight jaw muscles that they can only relax for about thirty seconds when directed to relax. In many of these cases you have to discount the beta at T3 and T4 and do your best with the rest of the record. The good news is that this is mostly a problem with eyes open and that it is less of a problem with eyes closed recording. Frontalis muscle tension from the forehead is another area where this occurs as well. An example of this can be seen in Fp1 and Fp2 of the record below.

As a consequence of the foregoing, the frontal leads are often contaminated with artifact, rendering them of minimal use. The temporal leads are often not much better. This can disqualify interpretations of Laplacian analysis and LORETA analysis as well. Marvin says he doesn't trust any beta in the record. Some individuals have so much muscle tension from shoulders, neck and jaw that it even migrates into the leads in the central region up into the vertex. It can also show up in the occipital leads from neck tension.

In spite of all these seemingly insurmountable obstacles, it is possible to get pretty good EEG recordings most of the time in the eyes closed condition. Clearly, for medical EEG technologists, this is a much bigger problem because they must have precise recordings for diagnosis. For neurofeedback therapists, it is less of a problem because we are training EEG in a more general manner, unless you are doing LORETA or Z score training. We are not doing medical diagnosis. When something is wrong with the EEG with respect to neurofeedback training, it is usually pretty obvious. This is why we were able to do neurofeedback so long without a qEEG. However, as our technologies become more refined and we engage in techniques such as live sLORETA training, as in the case of Brainmaster's Avatar technology, the importance of artifact free EEG will become critical.

According to E Roy John (John et al, 1987) and many others, the most reliable EEG for evaluation purposes is the eyes closed resting EEG. From the foregoing, it should be obvious at this point from a technical perspective why this is so. John actually did the math and the research to prove it. For this purpose, practitioners should rely on the eyes closed map for their primary analysis and review the eyes open map for any additional information that can contribute to the analysis. In both cases they should be vigilant for the distorting effects of artifact. Any doubt regarding the data in the map should drive them back to the raw EEG record for review. In order to review the raw data, they should know the basic difference between normal EEG, artifact and abnormal EEG. Note that I said "basic." Neurofeedback therapists should not try to be experts in medical EEG analysis. This requires years of study and training. It is outside our scope of practice. One of the best books on artifacting for neurofeedback therapists is the one published by ISNR and written by Cory Hammond and Jay Gunkleman (2011).

Scope of Practice Considerations

A chiropractor is not at fault for failing to spot a glioblastoma, brain tumor, in one of his or her patients. This kind of diagnostic skill is outside of his scope of practice. It is the same with neurofeedback therapists and brain tumors. Some practitioners would like to try and make it otherwise. If you are trained at a medical level and spot such a tumor that is great. It is unrealistic to expect other practitioners without medical training to do the same. Neurofeedback cuts across many disciplines and people with varied levels of skills and licenses utilize it. It is inappropriate for a chiropractor to treat mental illness beyond basic anxiety or depression that is metabolically based, such as hypothyroid driven depression, and especially personality disorders. A neurotherapist should have an idea of what seizure looks like but should not attempt to diagnose it. I have seen even well trained qEEG collectors get excited about misleading data- just ask Marvin. Since doing neurofeedback without a qEEG is not recommended by the ISNR panel of experts and a position paper has been published accordingly, and this author agrees, then most neurofeedback practitioners should strive to develop some level of skill in this area as it relates to their discipline. It should not mean, however, that we all become qEEG specialists. This author disagrees with the notion that we need another specialty certification beyond neurofeedback for qEEG guided neurofeedback. If qEEG should be part of neurofeedback, then it should be part of the training and the certification. Many will try to import neurofeedback into their exclusive medical discipline using contrary arguments. Others will try to make neurofeedback the domain of psychologists only. This will of course be argued endlessly in the future based on hidden political and financial agendas. I have found incompetence in all areas of professional practice. Neurofeedback requires a unique combination of skills that draws from several disciplines and must eventually become a field in its own right.

Editing & Editing Tools

To remove artifact from the raw EEG record, it is usually necessary to use a program that allows you to cut it out of the record. This typically involves visually inspecting the entire record in segments, often referred to as epochs, and marking and discarding segments that are contaminated. Below is a sample of the online artifacting tool made by New Mind Maps. You can see all 19 channels of EEG and segments without artifact that have been selected for keeping is highlighted in pink between 1:16 and 1:17. There are different signal processing methods that can be used to accomplish this, and the topic is far too technical to understand without specific training in this area and a good understanding of calculus. However, some basics can be presented in understandable terms.

Collecting the Edited Sample

In editing, one is trying to collect a representative sample of artifact free EEG. This term "representative" is important because that is how the databases were artifacted. You want a sample of everything that is happening. Theoretically, too much sample from one area of the record may not generate a good sample. In reality this does not always work out. Many people become drowsy after about three minutes of sitting with their eyes closed, especially if this is during the circadian dip in the afternoon and additionally if they did not get a good night's sleep the night before. Therefore, it is often best to get most of the sample in the first three minutes. One caveat related to this strategy is that people typically generate a burst of alpha when they first close their eyes for about 30-60 seconds that may bias the record. For this reason, it is a good idea to wait for a minute after the client closes their eyes before beginning the recording or begins editing a minute into the record.

In sampling one would want to avoid periods of drowsiness in the record and that should be obvious by sudden dropouts of alpha for long periods followed by increases in theta and then delta. Abnormal variants are fine as long as they are not related to sleep. Intermittent spikes or sharp waves are unlikely to distort the record significantly as shown below between 1:45 and 1:46.

Watch for eye movements in the frontal leads that result in false delta pattern. Eye blinks and eye rolls, as shown previously, should be obviously avoided. Neuroguide provides an ongoing split-half and test- retest measure to monitor for sample quality and state changes. This particularly enhances the precision of artifacting but is not necessary if one is trained properly and careful to avoid drowsiness in the record. As Thatcher often demonstrates by comparing an artifacted with a non artifacted record, in a good EEG sample artifacting does not change the picture that much. The moral of the story is again, garbage in- garbage out. Editing should only be done when it is absolutely necessary. The best EEG data is unedited data. Developing good skills at collecting data can avoid considerable editing. Editing is not a guarantee that you will get a good record. Below are two samples of the same data that have been edited. One is one minute of unedited data with artifact and the other is two minutes of edited data without artifacts. You can see that the edited version distorted the level of alpha in the map.

At times the abnormality that you are looking for may occur intermittently for several seconds. I often refer to these as abnormal recruitment cycles. They occur in ADHD, TBI, Sleep Disorders, and a variety of other rare problems. You may want to edit for a specific sample to be used exclusively for a separate map to better understand their frequency distribution and localize them more effectively using Laplacian and/or LORETA analysis.

Windowing

Windowing is an esoteric topic but for those interested in the details of how an EEG record is edited, especially for research, it can be a critical concern. Windowing involves taking separate unrelated selections of EEG and knitting them back together into one whole record. Some of the segments knitted together mathematically may be quite different and result in the addition of considerable noise and distortion into the record depending on the number of edits and the method used.

There are various editing platforms available and most of them are integrated into the databases available for EEG analysis or the qEEG amplifier programs presently on the market. The best form of editing is not editing. Editing artifact generates artifact, although usually subtle. The less editing the better. So, prevention is the key. Collecting artifact free EEG is the most accurate and efficient way to gather data.

The MiniQ and Editing

The New Mind MiniQ was collected using Brainmaster signal processing. The matrix was calculated based on typical levels of artifact generated by a user of this system. Using a more artifact free database system with this type of data would actually bias the matrix and the resultant maps. John Demos generated a primitive MiniQ database using Excel and Brainmaster summary excel files that worked quite well for many beginning users but had a limited output and processing capabilities. Thatcher provided a method to upload Brainmaster .eo files and edit them so they could be processed through his normative database. This enhanced accuracy but the report output was very limited. The New Mind system was designed to overcome these limitations and provide a high-quality output specifically for clinical use utilizing two or four channel amplifiers and sequential collection techniques. This allows individuals entering the field to have access to high quality database output and learn proper collection and map interpreting techniques before investing in high end research-oriented database systems such as NxLink, Neuroguide, or Sterman Kaiser.

Artifacts & Artifacting

This content is for customers only. Please purchase the course on the Shop page to view this content or log into your account if you have already purchased this course.