This article details simultaneous electroencephalogram and functional magnetic resonance imaging (EEG-fMRI) recording procedures that can be used in both clinical and research settings. EEG processing procedures to remove imaging artifacts for clinical review are also included. This study focuses on the example of epilepsy during the interictal period.
Simultaneous electroencephalogram and functional magnetic resonance imaging (EEG-fMRI) is a unique combined technique that provides synergy in the understanding and localization of seizure onset in epilepsy. However, reported experimental protocols for EEG-fMRI recordings fail to address details about conducting such procedures on epilepsy patients. In addition, these protocols are limited solely to research settings. To fill the gap between patient monitoring in an epilepsy monitoring unit (EMU) and conducting research with an epilepsy patient, we introduce a unique EEG-fMRI recording protocol of epilepsy during the interictal period. The use of an MR conditional electrode set, which can also be used in the EMU for a simultaneous scalp EEG and video recording, allows an easy transition of EEG recordings from the EMU to the scanning room for concurrent EEG-fMRI recordings. Details on the recording procedures using this specific MR conditional electrode set are provided. In addition, the study explains step-by-step EEG processing procedures to remove the imaging artifacts, which can then be used for clinical review. This experimental protocol promotes an amendment to the conventional EEG-fMRI recording for enhanced applicability in both clinical (i.e., EMU) and research settings. Furthermore, this protocol provides the potential to expand this modality to postictal EEG-fMRI recordings in the clinical setting.
Epilepsy afflicts nearly 70 million people worldwide1. As many as one out of every 150 people with poorly controlled epilepsy succumb to a sudden unexpected death in epilepsy (SUDEP) every year. Furthermore, roughly 30%-40% of epilepsy cases are refractory to medical management2. Neurosurgical treatment in the form of resection, disconnection, or neuromodulation can be a life-changing and life-saving measure for patients with refractory epilepsy.
Simultaneous electroencephalogram and functional magnetic resonance imaging (EEG-fMRI) is a unique combined technique that measures brain activities noninvasively, and has provided benefits for understanding and localizing seizure onset in epilepsy3,4,5,6. Scalp EEGs can be used to lateralize and localize seizure onset zones, but they have relatively poor spatial resolution due to limited capabilities with respect to assessing deep epileptogenic sources. While fMRI has good spatial resolution throughout the brain, including deep regions, fMRI alone is not specific to seizures. However, scalp EEGs can inform the interpretation of blood oxygen level-dependent (BOLD) activation or deactivation areas in fMRI, thereby producing an fMRI technique that is specific to epilepsy. Thus, the implementation of simultaneous EEG-fMRI can be used to map spatiotemporal processes relevant to localizing both the 'where' and 'when' of epileptic events.
Explanations of how to conduct simultaneous EEG-fMRI are provided in the previous studies7,8,9,10. However, EEG-fMRI has been underutilized in epilepsy, especially in clinical settings. There exists a study that provides a general procedure for EEG-fMRI recordings, background, and examples of possible EEG analysis7. Also, a study emphasizing hypnotic induction along with temperature measures in simultaneous EEG-fMRI recordings has been conducted8. Furthermore, an expanded EEG-fMRI study to introduce a spatiotemporal and fMRI-constrained EEG source imaging method has been proposed9,10. In addition, the use of a carbon wire loop to effectively remove artifacts from EEG-fMRI has been considered10. However, all these studies fail to address challenges in conducting EEG-fMRI studies in a clinical research setting. In particular, the use of the EEG cap restricts the feasibility of these protocols in clinical settings, and details about patient management are also missing. In this study, we provide an EEG-fMRI recording protocol that can be used in both clinical and research settings for patients with epilepsy. This unique protocol allows an easy patient transition from an epilepsy monitoring unit (EMU) to the scanning room. In addition, the protocol provides the potential to expand its application to postictal period recordings with epilepsy patients. For EEG-fMRI, post-processing is a crucial step in removing artifacts caused by MRI gradients and physiological artifacts, such as those related to heartbeat. Thus, we also provide step-by-step procedures to remove EEG artifacts using a standard template removal method11 for clinical review.
This study was performed in compliance with Protocol #62050, approved by the Institutional Review Board at the University of Kentucky (UK).
1. Subject recruitment
NOTE: Once the patient is admitted to the EMU, after the patient's agreement on the consent form to participate in this study, the following steps will be followed.
2. Electrode placement
3. EEG-fMRI recording
NOTE: For the EEG-fMRI recording, the MR conditional EEG recording system is used along with the MR conditional electrodes placed at the beginning of EMU admission.
4. EEG artifact removal
NOTE: The following descriptions provide detailed steps on how to remove the scanner artifacts on the EEG data obtained from the simultaneous EEG-fMRI recordings. Figure 6 displays the processing pipeline with representative examples.
When a patient is admitted to the EMU, a simultaneous scalp EEG and video are recorded. One purpose of the EEG monitoring by a neurologist is to evaluate for epileptiform discharges, which can potentially inform the localization of the seizure onset. When extraordinary dynamics of specific EEG channels are distinguished, the electrode locations can be associated with seizure localization. During the interictal period, interictal epileptiform discharges (IEDs), including spikes and sharp waves, are traditionally considered as markers of areas of epileptogenicity. Furthermore, the obtained recordings of interictal EEG-fMRI data can be advantageous to understand and localize the seizures. To validate this EEG-fMRI recording and EEG processing protocol, we evaluate the quality of the EEG by comparing the postprocessed EEG to the one from the EMU, and we confirm that the same distinguishable EEG characteristic is observed in both cases.
Focal delta slow activity is typically suggestive of an underlying brain structural lesion or an area where the neurons are not functioning as expected, which is often observed after strokes, head injuries, brain infections, or dementia. However, it is not uncommon for patients with focal epilepsy to demonstrate focal delta activity near or at the site of their epileptogenic zone. In addition, although focal delta activity is less specific than IEDs, it can specify localized structural pathology corresponding to seizure onset in epilepsy12. Furthermore, focal interictal slow waves in EEGs correspond to focal BOLD activation on fMRI that matches the epileptogenic area in patients with partial epilepsy13.
It is notable that, in temporal lobe onset epilepsy, a type of delta activity called temporal intermittent rhythmic delta (TIRDA) is sometimes present, and it is considered an IED equivalent14. On the contrary, temporal intermittent polymorphic delta (TIPDA) is not considered an IED equivalent15. In the processed EEG data, clear focal left temporal slow waves (TIPDA) are present, which is observed from the EEG recorded at the EMU (Figure 7). Although this delta activity is not IED equivalent, it suggests left temporal neuronal dysfunction.
Figure 1: An example of 32 electrode selections for simultaneous EEG-fMRI recording. In the clinic, 21 channels are commonly considered for EEG monitoring. To fulfill a minimum number of electrodes to conduct EEG source imaging (ESI), 11 additional channels are included to cover the head entirely. All electrodes are gold cups to avoid MR effects. In the figure, different colors are used to distinguish different electrodes, and the colors match the physical cable colors. Each rectangular box in the bottom indicates one junction to be connected to a harness, which will be connected to an amplifier for recording. Please click here to view a larger version of this figure.
Figure 2: Electrode placements. (A) Placement of the electrodes on the patient's scalp and (B) arrangement of the cable junctions. The left images in (A) and (B) provide a top-front view, and the right images provide a left-side view of the patient. The red arrows in (B) indicate the placement of gauze pads. This helps to avoid imaging artifacts. The areas circled in blue in (B) show how the cable junctions are arranged. Please click here to view a larger version of this figure.
Figure 3: Equipment connection in the monitoring room. (A) An overview of the USB 2 adapter, Syncbox, and Triggerbox connection. Detailed picture of (B) the USB 2 adapter and Syncbox cable connections, (C) the Syncbox and the fiber optic cable connection, and (D) the cable connections in the Triggerbox. The star marks on (B), (C), and (D) show the location of the USB cables to be connected to the recording PC. A schematic diagram of the EEG recording system and the connections required between the hardware is provided in Figure 1 in Mullinger et al.7. Please click here to view a larger version of this figure.
Figure 4: Equipment connection in the scanning room. (A) An overview of the EEG amplifier connection in the scanner. (B) Wrapped cables from the interface box to connect the EEG electrodes (the red cable is for the ECG measure). (C) Connection of the interface box and the EEG amplifier and MR safe sandbags placed to reduce MR artifacts. (D) Connection of the amplifier (top) and the battery (bottom) and connection of the fiber optic cable from the Syncbox in the monitoring room to the amplifier. Please click here to view a larger version of this figure.
Figure 5: Screenshot of workspace settings on the EEG recording software. The number of channels and sampling rate can be set under the amplifier settings. In addition, the specification of each channel can be modified, if necessary, by clicking the table at the bottom. Please click here to view a larger version of this figure.
Figure 6: EEG artifact removal pipeline with representative examples. The raw EEG traces are displayed in the bottom left. The bottom middle plot shows EEG traces after applying MR artifact correction and a high pass filter of 0.5 Hz on the raw EEG. The bottom right plot displays EEG traces after applying CB artifact correction and a notch filter of 60 Hz on the processed EEG. The EEG traces are displayed under common ground mode to effectively visualize how each process influences each recorded channel. Please click here to view a larger version of this figure.
Figure 7: Comparison of the processed EEG from a simultaneous EEG-fMRI recording (left) and an EEG recorded at the EMU (right). The red circle indicates focal left temporal slow waves at the same channels. The EEG traces are displayed in a double banana format, which is traditionally considered in the clinic. Please click here to view a larger version of this figure.
This experimental protocol is unique in providing a smooth transition of patients with epilepsy from the EMU to the scanning room, allowing it to be used in clinical and research settings. The use of FDA-approved MR conditional electrodes is an essential component for both clinical recordings during the time spent in the EMU and for safe transfer to MRI without having to remove or exchange the scalp electrodes from the patient. In the EMU, the MR conditional electrodes are connected to an amplifier for simultaneous video and EEG monitoring. For EEG-fMRI recordings, an MR conditional EEG amplifier and an MRI scanner can be used with a 20-channel head-coil, which accommodates the size of the electrode set and connecting wires. It must be noted that before conducting the simultaneous EEG-fMRI recordings in patients with epilepsy, a test run with a healthy subject is highly recommended to confirm the proper operation of all equipment and to become familiar with each required step.
In addition, concrete organization of the team and careful selection of patients also play a significant role in this protocol. To be viable for both clinical and research settings, it is required to have a structured team of epileptologists, nursing staff, EEG technologists, and engineers. For patient selection, the above-listed inclusion and exclusion criteria must be firmly considered.
Furthermore, it is important to address that when EEG-informed fMRI analysis is conducted, clear presence of the key features of EEGs must exist to guide the corresponding BOLD changes in fMRI. Therefore, when conducting the EEG-fMRI recording, it is important to consider patients who have previously demonstrated target EEG features. During the interictal period in patients with epilepsy, IEDs, which are abnormal and suggest epileptogenic potential, are a well-known EEG feature to reference to the BOLD changes16, even though the example here does not include this case. When targeting to obtain IEDs in the interictal EEG-fMRI recordings, experimenters should consider patients with frequent IEDs (at least three IEDs/hour) observed by a scalp EEG, in order to ensure sufficient epileptiform discharges during a scanning session. The number of IEDs can be determined from the EEG monitoring in the EMU, or from referencing the IED frequency seen in the subjects' prior EEG recordings if they exist. The obtained recordings of interictal EEG-fMRI data can bring benefits to understanding and potentially localizing the seizure onset zone17.
Once a clean EEG is obtained after processing the artifact removal steps, further EEG analysis can be applied. For example, EEG source imaging (ESI) can be obtained by applying standardized low-resolution brain electromagnetic tomography (sLORETA)18 to estimate the brain's corresponding electrical activity on the cortical surface. The estimated sources can be obtained by inverting the computed lead field matrix based on the head, outer skull, inner skull, and cortex layers created from the patient's MRI using the boundary element method19. There are numerous publicly available toolboxes to obtain EEG source imaging, and Brainstorm is one popularly used MATLAB-based toolbox20.
When ESI is considered using the processed EEG, the total number of electrodes and their distributions must be carefully taken into account so that they can reasonably cover the entire head. The minimum number of electrodes necessary to implement ESI is 32 channels21,22, which is more than the standard number of electrodes used in clinical settings. Thus, it is recommended to include extra channels to cover the entire head with reasonable spacing. The channel selection in this study includes 21 channels, which are conventionally used in the clinic for EEG monitoring, and 11 additional channels to cover the head entirely (Figure 1).
Here, we do not include details of fMRI analysis, since this is out of the scope of our study. However, a possible direction is EEG-informed fMRI analysis23. For instance, the occurrence time of IEDs can be saved as event triggers to correlate with the fMRI, which can lead to a routine event-related fMRI analysis. In this case, a generalized linear model analysis can be used to find the brain regions showing changes in fMRI signal at the time of IEDs.
We point out that a recently published study10 has shown it is possible to use a carbon wire loop system when a more robust artifact removal technique is required16. However, we want to apprise that the integration of the carbon wire loop system in our experimental setting with the MR conditional electrode has not been investigated yet.
Even though this study specifically focuses on the interictal period of epilepsy, the introduced protocol for simultaneous EEG-fMRI can be further extended to the ictal or postictal period. However, specific considerations must be followed when any customized settings are considered. For the postictal phase, an important concern that we are cognizant of is that the patient is given a benzodiazepine prior to transport to the MRI. As for the frequency analysis of EEGs, it has been reported that benzodiazepines do not necessarily alter the specific frequency bands24,25, and in the case of modest changes, these are confined to the somatosensory-motor region26 or frontal lobes27. Furthermore, with respect to simultaneous EEG-fMRI, delta EEG-BOLD correlations showed no changes after benzodiazepine injection compared to a control with saline injection27. The BOLD signal was decreased in only the small areas of Heschel's gyrus and supplementary motor area.
The authors have nothing to disclose.
This work was partially supported by the College of Medicine, the Vice President for Research, UK HealthCare, and the Research Priority Area at the University of Kentucky as part of the College of Medicine Alliance Initiative and Dr. Jihye Bae's Start-Up funds provided by the Department of Electrical and Computer Engineering at the University of Kentucky. The authors thank the volunteer participants for the recording and the Epilepsy-Neuroimaging Research Alliance team members, especially Dr. Brian Gold for leading the alliance team, Dr. Sridhar Sunderam for research mentorship, and Susan V. Hollar and Emily Ashcraft for patient care and management.
3T Magnetom Prisma fit MRI scanner | Siemens Healthineers | ||
Abralyt HiCl, 10 g. | EASYCAP GmbH | Conductive gel for ECG electrode. | |
BrainAmp MR plus 32-channel | Brain Products GmbH | S-BP-01300 | |
BrainVision Analyzer Version 2.2.0.7383 | Brain Products GmbH | EEG analysis software. | |
BrainVision Interface Box 32 inputs | Ives EEG Solutions, LLC | BVI-32 | |
BrainVision Recorder License with dongle | Brain Products GmbH | S-BP-170-3000 | |
BrainVision Recorder Version 1.23.0003 | Brain Products GmbH | EEG recording software. | |
Collodion (non-flexible) | Mavidon | Glue to secure EEG electrodes. | |
Fiber Optic cable (30m one line) | Brain Products GmbH | S-BP-345-3020 | |
Gold Cup Electrode set, 32 channel | Ives EEG Solutions, LLC | GCE-32 | 2+ items are recommended when managing multiple subjects with overlapped/close period of Epilepsy Monitoring Unit (EMU) stay. |
Gold Cup Electrodes | Ives EEG Solutions, LLC | GCE-EKG | |
Harness, 32 lead, reusable | Ives EEG Solutions, LLC | HAR-32 | 2+ items are recommended when managing multiple subjects with overlapped/close period of Epilepsy Monitoring Unit (EMU) stay. |
MR-sled kit including 100% and 75% length base plates, low profile (3 cm) block legs for each base plate, ramp, and strap systems as hand configured | Brain Products GmbH | BV-79123-PRISMA SKYRA | |
Natus NeuroWorks EEG | Natus | Software used for EEG monitoring at the Epilepsy Monitoring Unit (EMU). | |
Nuprep Skin Prep Gel | Weaver and Co. | ||
Passive starter set, including consumables (gel, syringes, dispensing tips, adhesive washers, etc.) to facilitate out of the box data acquisition | Brain Products GmbH | S-C-5303 | |
SyncBox compl. Extension box for phase sync recordings | Brain Products GmbH | S-BP-02675 | Syncbox |
syngo MR XA30 | Siemens Healthineers | Software used for the MRI scanner. | |
Ten 20 Conductive Neurodiagnostic Electrode Paste | Weaver and Co. | Conductive gel for EEG electrodes. | |
TriggerBox Kit for BrainAmp | Brain Products GmbH | S-BP-110-9010 | Triggerbox; This Kit allows to expand the trigger width from the scanner so that the trigger signal can be detected on the BrainVision Recorder properly. This kit may not be required depending on the characteristics of the trigger signal provided by the scanner. |
Xltek EMU40EX amplifier | Natus | An amplifier used at the Epilepsy Monitoring Unit (EMU). |