New Record(ings)!
Using iEEG to record and analyze brain signals
Welcome to Brain Blast, your weekly deep-dive into the worlds of brain science, cognition, neuroscience, and psychology. Today, we're talking about intra-cranial electroencephalography, or iEEG, which uses tiny electrodes implanted in or on the brain to collect electrical signals from the surrounding neural populations over time. This is an incredible technology with a rich history that has served as a critical backbone for neuroscience research.
Click on the headings below to jump straight to a section!
In The News: Precision Neuroscience breaks world record in iEEG recording
Histories and Legacies: iEEG, epilepsy, and Dr. Wilder Penfield
Review Corner: Using iEEG to explore attentional control
In The News
On May 28, 2024, Precision Neuroscience announced to the world that it had set a world record in April of that year for the "number of electrodes placed on and recording cortical data from a human brain", as quoted in a statement published at GlobalNewsWire. The record of 4,096 electrodes doubled the previous record of 2,048 set by Tan and colleagues in their 2023 publication in the Journal of Neurosurgery. While much of the news cycle fixated on Precision's role as a competitor to Neuralink, an Elon Musk company, considerably little focus was given to just how impressive and cool this record is, and exactly what hurdles Precision has overcome to achieve it.
Precision's recording device uses a tried and true design, which embeds microscopic electrodes in a square grid pattern in a paper-thin microfilm that sits on top of the brain. The number recordings, then, is simply the number of electrodes placed in the grid. So why not just add more electrodes? The answer lies in the brain's anatomy. Imagine you have a 1-inch piece of paper and are asked to lay it flat on your forehead. Now imagine doing the same thing with a 2-inch piece of paper. Now, 3 or 4. Eventually, the paper becomes large enough that it has to fold to lay "flat" on your head. The same problem emerges as recording microfilms become larger - eventually, they can no longer lay on the surface evenly to record.
Tan and colleagues reported an ingenious solution to this issue, which led to their record-holding stint in 2023. Rather than recording from larger and larger grids, Tan and colleagues simply added separate grids, laid side by side. In this way, they could double the number of recording sites while keeping the grid size small enough to conform to the changing contours of the brain's surface. While Precision's exact methodology has not been published to date, their website specifically touts their device's "modular" design, which is probably the same idea. In fact, their electrode grid has the same number of electrodes - 1,024 - as Tan and colleagues'; they just used four grids instead of two.
While the number of electrodes that Precision recorded from is, on paper, simply an extension of Tan and colleague's original insight, there's more to this development than meets the eye. Recall the 1-inch paper you imagined laying on your forehead earlier. This is approximately the size of the grid used by Tan and colleagues, which was 32x32mm. In contrast, Precision's grid is just 16x16mm, cramming twice as many electrodes in in each direction. This huge leap in density allows Precision to collect much more data per unit area, and therefore have much more precision (heh) in localizing signals to different areas on the brain.
It's worth noting that Precision has achieved this milestone while simultaneously developing novel surgical methods that make insertion of their device less invasive than previous techniques. As illustrated in Figure 2H of Tan and colleagues' 2023 report (Warning: Graphic), microfilm-based iEEG devices are typically implanted on the brain's surface by exposing a significant area of the brain, on the order of several square centimeters. Precision Neuroscience reports that they have developed a novel method of insertion that they call a "cranial micro-slit" technique. While again, exact methods are not yet published for this technique, marketing materials from Precision show a simple procedure in which a small linear opening is cut into the skull through which the grid is inserted. While it is yet unclear whether this technique was employed when Precision conducted its already impressive record(ings), both are in any case exciting developments in iEEG techniques, and Precision Neuroscience should be commended for its achievements.
Histories and Legacies
Note: Basic details on Wilder Penfield's life and epilepsy treatment are generally synthesized from Wikipedia’s Wilder Penfield and Epilepsy pages as well as The Neuro at McGill
Most people are vaguely aware that brain surgery can be and is often conducted on awake patients, but perhaps fewer know why. In fact, this is so the surgeon can work with the patient to identify and preserve areas of the brain that serve critical functions, such as motor and language generation. This process, called "awake craniotomy", was originally developed and still serves as a critical basis for surgical treatment of epilepsy. Interestingly, awake craniotomy and its use in epilepsy were critical for the development and proliferation of iEEG as a cornerstone technique of brain research, and it's thanks to a man named Wilder Penfield.
Wilder Graves Penfield was born in 1891 in Spokane, Washington in the United States. After attending Princeton University for his undergraduate degree, he received a Rhodes Scholarship (which, fun fact, had been established in just 1902) to study medicine at Oxford. During this time, he was introduced to brain science by Sir Charles Sherrington, who, among other things, coined the term "synapse", but was soon pulled away as a result of World War I. Eventually, he completed his medical degree at Johns Hopkins University and went briefly into surgery before returning to Oxford to complete his studies there in 1921. He continued to study under the tutelage of several prominent researchers in Spain and New York, in particular learning about epilepsy and related surgical techniques, before moving to Montreal, Quebec in Canada and becoming an established neurosurgeon at the Royal Victoria Hospital. There, he established the Montreal Neurological Institute (MNI), using funds from an endowment by David Rockefeller (yes, of THE Rockefellers).
One of Dr. Penfield's critical contributions to neuroscience was the "Montreal Procedure", a detailed multi-step process for approaching surgical epilepsy treatment that he pioneered with Dr. Herbert Jasper. Epilepsy is characterized by abnormal spurts of neural activity, which are often initialized in particular source locations, or "foci", in the brain. In particular, Penfield and Jasper outlined a series of presurgical steps that aimed to identify the minimal area of the brain that would need to be removed (or "resected") in order to effectively treat the epilepsy without otherwise unduly impacting brain function. A critical component of these steps was the use of iEEG (then, and still often, referred to as electrocorticography, or "ECoG") to record candidate areas for resection to confirm the presence of epileptic spiking activity.
Penfield oversaw 1,132 surgeries following the Montreal Procedure as director of the MNI. Over time, findings from Penfield and other surgeons led to the development of the classic temporal lobectomy procedure. This procedure removed several structures of the medial temporal lobe (MTL; if you flipped your brain over on its top, it would be the inner portion of the two segments of cortex that are separated by the subcortical areas and brainstem), which were often associated with epilepsy. While this procedure significantly improved surgical outcomes, it had several unintended side effects (For example, avid neuroscience readers will be familiar with H.M., or Henry Molaison, who became a cornerstone case for our understanding of memory formation after he developed severe anterograde amnesia following a particularly extensive bilateral MTL resection). Since then, further developments in our understanding of the brain and the underpinnings of epilepsy, as well as improvements in iEEG recording, have further refined surgical interventions, leading to even better surgical outcomes while vastly improving postsurgical side-effects.
Dr. Penfield passed away in Montreal in 1976, but his legacy is still strong in neuroscience tradition. Today, iEEG remains a critical component of surgical epilepsy intervention, and indeed epilepsy patients represent a large portion of subjects who have participated in basic science efforts that only those patients undergoing awake craniotomies can take part in. The MNI still stands as a top-tier institute for brain research, both in basic science and in clinical application efforts. Penfield himself was designated as a National Historic Person of Canada in 1988 and, 20 years later, was awarded the honor of being the subject of a Google Doodle.
Review Corner
Today's article, "Attentional control influence(sic) habituation through modulation of connectivity patterns within the prefrontal cortex; Insights from stereo EEG" was published open-source in the journal Neuroimage on May 7, 2024. It is a primary research report by Huimin Huang, Rui Li, Xiaojun Qiao, Xiaoran Li, Siyi Chen, Yi Yao, Fengpeng Wang, Xiaobin Zhang, Kaomin Lin, and Junsong Zhang, in affiliation with Central China Normal University, Xiamen University, and Xiamen Humanity Hospital.
Our article today uses a different variant of iEEG than the 2-dimensional surface microfilms we've talked about above, called “stereo EEG”. In stereo EEG, electrodes are arranged along a thin, long rod and inserted deep into the brain's tissue lengthwise. This method is referred to as "stereo" because of its use of stereotaxic localization to guide the insertion of the electrodes toward the intended target. This technique works much like a car GPS system: a triangulation camera tracks the position of the electrodes, like satellites track the position of your car, and that position is aligned to the patient's brain scan the way your car’s position would be referenced to a map. Because the electrodes are on inserted rods, we can collect data from deeper brain structures this way.
Patients in this study underwent iEEG recording while they performed a classic Stroop task. If you don't remember this one from class, it's where you have to read the color of the ink a word is written in, and the words are also the names of colors, but not necessarily the same color as the ink. The critical manipulation in a Stroop task is called "congruency" - whether the ink color and the word are the same (congruent trials) or not (incongruent trials). In general, participants will respond more slowly on incongruent trials than on congruent ones; this is called the "Stroop Effect".
The base psychological concept we talk about when discussing the Stroop Effect is called "control". In general, control refers to the set of processes that guide our behavior as a function of our current environment and internal motivations. In the Stroop task, it's the processes that direct your attention to the words and their ink colors, interpret them, and then select and implement your response. Note that control is involved both for processing the task-related response (i.e., choosing a response based on the color of the ink) and for processing the automatic response (i.e., choosing a response based on the word you are reading). The response you give on any particular trial is the sum total result of these competing processes.
This latter point is a key aspect of how the brain controls our behavior: different areas of the brain are responsible for different control processes, and yet other areas are responsible for integrating these different processes into a cohesive response. Several of these regions seem to be located in the prefrontal cortex, especially those involved in integration and response selection. To intuit where the prefrontal cortex is, put your hands up by your ears like you're imitating an elephant, then move them forward to your temples. The prefrontal cortex is the part of your frontal lobe that extends forward from roughly that point.
The authors' goal in this study was to use iEEG to sort out the distinct processes being contributed by different regions of the prefrontal cortex during performance of the Stroop task. To understand how they did this, it's important to understand the kind of data that iEEG can give us. Neurons produce electrical signals through shifts in ions that cause voltage changes in the space around the neurons. The iEEG signal is a measure of the aggregate voltage of that space over time. The resulting data, called a "time series", is pseudosinusoidal, which means we can analyze it in both the time and frequency domains.
With this data, the authors identified three areas of prefrontal cortex that showed neural activity in response to participants actively doing the Stroop task: the inferior frontal gyrus (IFG); the middle frontal gyrus (MFG), and the orbitofrontal cortex (OFC). In our previous demonstration of the location of the prefrontal cortex, the IFG is the area just next to your first knuckle, the MFG is by your second knuckle, and the OFC is toward your forehead (Figure 1C gives a good visual). Next, the authors aimed to demonstrate how they interacted with one another in the service of attentional control.
I’ll highlight two analyses they used to do this: Granger causality and cross-frequency coupling. Granger causality looks at how the activity of one region may be driving activity in another region (i.e., how a region's neurons firing "causes" another region's neurons to fire). In a simple example, imagine we had elecrodes recording from just two neurons. If the firing of neuron A causes the firing of neuron B, then we would see one spike from each electrode, where the spike from neuron A would show up just a bit earlier than the one from neuron B. Granger causality analysis identifies this association mathematically in much more complex data, but the principle is the same. In this case, the authors found that causality flowed from the OFC to the MFG and then to the IFG, forming a hierarchy of information flow across the three regions.
Cross-frequency coupling measures how the brain is coordinating activity across multiple regions. As I mentioned above, the time series collected in iEEG is pseudosinusoidal, and you might remember from math class that sinusoidal signals have a phase, an amplitude, and a frequency. To coordinate activity across multiple regions, the brain links the amplitude of each region's individual high-frequency activity to the phase of a low-frequency signal, called an “envelope”, so that each region's activity is changing in sync with each other (for a visual, see the lower left figure in this image). Cross-frequency coupling assesses if and which frequencies are involved in this coordination for a given set of regions. Here, coupling was found between the MFG and both the OFC and IFG, but with different low-frequency envelopes, suggesting that the MFG serves as an integrative center for two different communication streams.
This effect was further influenced by trial congruency, with congruent trials showing larger coupling between MFG and IFG and incongruent trials showing larger coupling between OFC and MFG. This suggests that the IFG and OFC are supporting separate control processes that relate to the type of control engaged by each type of trial congruency. More specifically, on congruent trials, participants could rely on the automatic response, presumably processed by the IFG, to make their response; in contrast, on incongruent trials, they had to overcome this automatic response with task-related information, ostensibly processed by the OFC.
As far as my review is concerned, I generally like the ideas this study plays with. iEEG studies of control processes are incredibly difficult – patient work is hard in the best of scenarios, and awake craniotomy even more so. The analyses undertaken in this study (of which I’ve only been able to describe a fraction here) are a pretty comprehensive breakdown of the ways we can leverage iEEG data to reveal information about the brain areas involved in task processing and how they communicate with each other. In general, I think the findings are quite interesting, and largely match what we would have expected based on previous work with less direct data collection methods like MRI.
That said, I would say that this paper suffers from a lack of clear connection between the authors’ reported results and their stated conclusions. In several instances, the results of a given analysis are reported numerically, but are never interpreted in any meaningful way, leaving the reader to figure out what they are supposed to be appreciating about that finding (a good example is the Granger causality analysis, which is presented on Figure 4b with no explanation given for how to interpret the figure). There are also enough grammatical errors present in the paper that it causes confusion in some places1; I am still not certain if the authors meant to use the words “congruent” and “incongruent” in the places they did in paragraph 5 of the discussion, because by my reading I think their proposed mechanisms are swapped. Additionally, I’m simply not convinced that the clean, two-pattern conclusion they present in paragraph 1 of the discussion really summarizes what the results taken together are showing us, and I’m not sure if that is because I think the results themselves are not as clean as the conclusion suggests, or if the authors simply have not made the connections between those results clear enough to get the reader to that conclusion.
Ultimately, as a reviewer I would probably have requested minor revisions on this paper to clarify my interpretation concerns above, but overall the study is sound and the results are impressive on their own. I’m always in favor of studies that highlight the distinct subregions of prefrontal cortex, and the replication of common MRI findings with iEEG is a strong confirmatory result.
I’ll note here that grammatical errors in general are not something I am terribly concerned with in academic papers. Publish or perish is real, and academics are not copy-editors, not to mention the very real issues with language barriers for international academics. The issue here is that these errors cause enough issues with interpretation that they call into question factual aspects of the report itself.