方法文章

药物难治性癫痫患儿术前评估中的电磁源成像

DOI：

10.3791/66494

⸱

2024年9月20日

 ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  , 
1Neuroscience Research Center, Jane and John Justin Institute for Mind Health, Cook Children’s Health Care System, 2Department of Bioengineering, University of Texas at Arlington, 3Burnett School of Medicine, Texas Christian University

本文内容

摘要

$$\rightleftharpoonup{xx}$$ $$\longleftharp{xx}$$, $$\longrightharp{xx}$$,

脑磁图（MEG）与高密度脑电图（HD-EEG）很少被同时记录，尽管二者可提供相互验证且互补的信息。本文展示了同步记录MEG与HD-EEG的实验设置，以及分析这些数据的方法，旨在定位药物难治性癫痫患儿的致痫灶和功能重要脑区。

摘要

$$\rightleftharpoonup{xx}$$ $$\longleftharp{xx}$$, $$\longrightharp{xx}$$,

对于药物难治性癫痫（DRE）患儿，实现无发作依赖于精确定位并切除（或损毁/断开）癫痫发生区（EZ），同时保护功能重要脑区。因此，开发一种可靠且无创的定位方法，为临床提供可用于定位EZ的有用信息，对于取得成功的手术效果至关重要。电生理源成像（ESI）和磁源成像（MSI）在这些患者的术前评估中应用日益广泛，已在癫痫发生区及功能重要脑区的定位方面展现出良好前景。此外，将ESI与MSI整合为单一解决方案，即电磁源成像（EMSI），基于同步的高密度脑电图（HD-EEG）和脑磁图（MEG）记录，其源定位准确性已证明高于任一单独模态。尽管这些结果令人鼓舞，但此类技术目前仅在少数三级癫痫中心开展，同步记录尤为罕见，在儿科人群中的应用仍显不足。本研究展示了同步采集MEG与HD-EEG数据的实验设置，以及分析这些数据的方法框架，旨在定位DRE患儿的激惹区、发作起始区及功能重要脑区。更具体而言，本文介绍了以下实验设置：（i）在睡眠期间记录并定位发作间期和发作期癫痫样放电活动；（ii）在视觉运动任务以及听觉和躯体感觉刺激过程中，记录视觉、运动、听觉和躯体感觉诱发电位，并绘制相应的功能重要脑区（即视觉、运动、听觉和躯体感觉区）图谱。此外，详细阐述了数据分析流程，包括采用等效电流偶极子（ECD）和动态统计参数映射（dSPM）方法进行EMSI以及单独的ESI和MSI分析的具体步骤。

引言

$$\rightleftharpoonup{xx}$$ $$\longleftharp{xx}$$, $$\longrightharp{xx}$$,

癫痫是最常见且致残性最强的神经系统疾病之一，其特征是反复发作且无明显诱因的癫痫发作，发作形式可为局灶性或全面性。尽管已有多种有效的药物治疗手段（例如抗癫痫发作药物[ASMs]），但约20-30%的患者无法控制发作，属于药物难治性癫痫（DRE）1。对于这些患者，癫痫手术是消除癫痫发作最有效的治疗方法；若能完全切除（或损毁/断开）致痫区（EZ）——即产生癫痫发作所必不可少的最小脑区——则手术可能取得成功2。在保留功能关键皮层的前提下，准确界定并切除（或损毁/断开）致痫区，是实现术后无发作的关键因素。为了确定是否适合手术，多学科团队会使用多种无创诊断工具来界定不同的皮层区域（即激惹区、发作起始区[SOZ]、功能缺损区和致痫性病变），这些区域可作为致痫区的间接近似指标3。当上述方法均无法明确识别致痫区时，则需要通过颅内脑电图（iEEG）进行术外监测。iEEG的作用在于精确定位发作起始区（即临床癫痫发作起源的脑区），从而明确致痫区位置，并绘制功能关键脑区图谱。然而，由于该方法具有侵入性4,5,6、空间覆盖范围有限，且需要事先具备清晰的术前定位假设7，因此存在严重局限性。结果可能导致未能发现真正的发作焦点及其实际范围，从而导致手术失败。此外，iEEG的解读需要在住院期间记录多次典型的临床发作，这延长了住院时间，增加了并发症风险（如感染和/或出血）5。因此，迫切需要开发可靠且无创的定位方法，以提供具有临床价值的信息，整体改善药物难治性癫痫患儿的术前评估。

近几十年来，电和磁源成像（ESI 和 MSI）在药物难治性癫痫（DRE）患者术前评估中的应用日益广泛，用于界定致痫区以及大脑功能区。特别是，ESI 和 MSI 可通过无创记录手段（如高密度脑电图（HD-EEG）和脑磁图（MEG））重建神经源活动，从而辅助手术方案制定或颅内脑电（iEEG）电极的放置。ESI 和 MSI 可用于定位发作间期癫痫样放电（IEDs），例如棘波和尖波，也可用于定位发作期（癫痫发作）活动。此外，这些技术还可用于定位参与感觉、运动、听觉和认知功能的不同功能脑区。通过对电生理事件（如 IEDs 和癫痫发作）进行重建，可分别识别激惹区（即 IEDs 起源的脑区）和癫痫发作起始区（SOZ），这两者均被视为定位致痫区（EZ）的有效替代指标。对功能皮层（即对特定皮层功能不可或缺的脑区）3的定位则有助于绘制功能区相对于计划切除范围的位置和范围，从而预先降低癫痫手术可能带来的功能缺损风险8,9,10,11。多项研究探讨了 ESI 和/或 MSI 在癫痫术前评估中的临床效用，结果表明这些技术在致痫区定位方面具有良好的应用前景12,13,14,15,16,17,18,19。例如，Mouthaan 等人14基于 11 项前瞻性与回顾性癫痫研究的无创数据进行了广泛的荟萃分析，报告称这些源定位技术总体上可高灵敏度（82%）但低特异性（53%）地识别致痫区。其他研究也表明，在磁共振成像（MRI）结果正常的癫痫患者中，MSI 和 ESI 能够准确地将癫痫灶定位在术后切除范围内19,20,21。这些定位结果对于因临床或影像学检查结果不明确而无法接受癫痫手术的患者尤为重要。综上所述，ESI 和 MSI 在 DRE 患者致痫区及功能脑区的术前定位中具有重要价值。

尽管这些发现令人鼓舞，但目前仅有少数三级癫痫中心常规开展此类技术，且在儿科人群中往往未得到充分利用。此外，高密度脑电图（HD-EEG）与脑磁图（MEG）很少被同步记录，而两者实际上可提供相互验证且互补的信息。MEG 对位于脑回表面、具有切向方向的浅表源敏感，但对径向方向的源或位于脑回深部区域的源则不敏感。22,23,24,25,26此外，与脑电图（EEG）相比，脑磁图（MEG）具有更高的空间分辨率（毫米级）。16,22,25与脑电图信号不同，脑磁图信号无需参考电极，且基本不受脑组织不同导电性（即脑膜、脑脊液、颅骨和头皮）的影响。25,27 提供大脑产生的磁场的无失真测量。另一方面，脑电图（EEG）能够检测所有方向的源信号，但其空间分辨率低于脑磁图（MEG），且更容易受到伪迹干扰26,28由于对源定位和深度的敏感性互补，约30%的癫痫样放电活动（如痫样放电，IEDs）只能在脑磁图（MEG）中记录到，而无法在脑电图（EEG）中检测到，且 反之亦然26,29,30,31,32与允许长时间记录的脑电图（EEG）不同，由于脑磁图（MEG）的记录时间受限，通常难以捕捉到临床发作事件，因为大多数患者的发作期事件无法在有限的记录时间内出现。此外，与发作相关的眼动或头部运动所引起的伪迹常常会干扰MEG记录的质量。29,33,34,35另一方面，与脑电图（EEG）相比，脑磁图（MEG）记录速度更快、操作更简便，尤其适用于儿童，因为无需在儿童头部安放传感器。35.

硬件技术的进步使得能够同时记录覆盖整个头部的高密度脑磁图（MEG）和高密度脑电图（HD-EEG）数据，传感器数量超过550个。此外，现代脑电图技术的发展已将高密度脑电图的准备时间缩短至不到15分钟36。这对于行为配合困难、无法长时间保持静止的儿科人群尤为重要。同时，软件技术的进步使得脑电图源成像（ESI）与脑磁图源成像（MSI）能够整合为一种联合解决方案，即电磁源成像（EMSI），该方法基于同步采集的高密度脑电图与脑磁图数据进行分析。多项理论与实证研究表明，EMSI在源定位准确性方面优于单独使用任一模态13,30,31,37,38,39,40,41。Sharon 等人.37采用不同的源定位方法重建对感觉刺激的脑活动，发现与单独使用ESI或MSI相比，EMSI的定位结果始终更接近功能性磁共振成像 (fMRI)——后者作为无创性精确定位准确性的参考标准。作者认为，这种定位精度的提升源于用于求解逆问题的传感器数量增加，以及两种成像模态不同的敏感性模式37。类似地，Yoshinaga 等人.31对局灶性难治性癫痫患者同步采集的脑电图和脑磁图数据进行偶极子分析，发现EMSI提供了单独使用任一模态无法获得的信息，并成功指导其中一名患者的癫痫手术定位。在一项前瞻性盲法研究中，Duez 等人.13显示，与单独使用ESI或MSI相比，EMSI显著提高了无发作概率（即更高的优势比），定位准确性达到或超过52%，与激惹区和癫痫发作起始区（SOZ）的一致性分别达到或超过53%和36%。我们团队最近的一项研究42表明，EMSI在定位估计精度和手术预后预测性能方面均优于单独使用ESI或MSI，其相对于切除区域和SOZ的定位误差分别约为8 mm和15 mm。尽管这些结果令人鼓舞，但目前仍缺乏针对药物难治性癫痫（DRE）儿童人群的EMSI方法学研究框架。

本研究展示了同步进行脑磁图（MEG）和高密度脑电图（HD-EEG）记录的实验设置，以及分析这些数据的方法学框架，旨在定位药物难治性癫痫（DRE）儿童的癫痫样放电区、癫痫发作起始区（SOZ）和功能重要脑区。具体而言，本文介绍了以下实验设置：（i）在睡眠状态下记录并定位发作间期和发作期癫痫样活动；（ii）在视觉运动任务以及听觉和躯体感觉刺激过程中，记录视觉、运动、听觉和躯体感觉诱发电位，并绘制相应的功能重要脑区（即视觉、运动、听觉和躯体感觉区）的功能图谱。此外，本文还详细阐述了数据处理流程，包括进行电磁源成像（EMSI）以及利用等效电流偶极子（ECD）和动态统计参数映射（dSPM）方法进行个体化脑电图源成像（ESI）和脑磁图源成像（MSI）的具体步骤。

方案

$$\rightleftharpoonup{xx}$$ $$\longleftharp{xx}$$, $$\longrightharp{xx}$$,

The experimental procedures applied here have been approved by the North Texas Regional Institutional Review Board (2019-166; Principal Investigator: Christos Papadelis). The following section will describe the experimental protocol for the noninvasive source localization of IEDs, ictal onsets, and event-evoked responses (i.e., visual, motor, auditory, and somatosensory) using simultaneous MEG and HD-EEG recordings followed in our laboratory. The International Federation of Clinical Neurophysiology43 and the American Clinical MEG Society44 have provided "minimum standards" for the routine clinical recording and analysis of spontaneous MEG and EEG data. Procedures for HD-EEG recordings described here apply only to sponge-based EEG electrode systems. The overall preparation process for each subject is about 2-3 h, comprising the actual recordings of ~1.5 h.

1. MEG system's preparation

  1. Before the subject's arrival, perform an empty-room MEG recording of a few minutes to capture the background noise levels and magnetic artifacts and check that all MEG sensors are working properly.
  2. Using the MEG sensors tuning program, ensure that the average white noise value of all MEG sensors is between 2 and 5 fT/√Hz (fT/cm√Hz for gradiometers).

2. Subject's preparation

  1. Ensure that the subject is comfortable with the environment. In the case of young children, allow them to explore the recording room (including the magnetically shielded room [MSR]) and see the testing equipment that will be used for data acquisition.
    1. Screen and provide the subject instructions using the screening consent form. If needed, explain the procedure to young children using special words, toys, and games developed for each age group. Ask the subject (or the subject's parents) if he/she had a seizure within the last ~2 h prior to the visit.
      NOTE: The screening consent form includes a description of the test, as well as its safety, why the testing is performed, and an overall description of the study.
  2. Remove any metallic and/or magnetic materials from the subject and provide the subject with appropriate hospital-issued clothes (e.g., hospital gowns, scrubs). In addition, ask the subject to remove his/her shoes to prevent magnetic dust from entering the MSR. If other ferromagnetic elements, such as dental works or implanted medical devices, cannot be removed, use a degausser (i.e., demagnetizer) to remove residual magnetic artifacts that can cause either interferences or high noise levels during MEG recordings. After making sure that all sources of magnetic noise have been removed, ask the subject to sit and get comfortable on a wooden chair where the next measurement procedures will be applied.
    NOTE: The degausser must not be applied directly to any implanted electronic devices (e.g., pacemakers, neuromodulation devices).
  3. Measure the subject's head circumference to select the appropriate EEG net size (usually 32-34 cm up to 58-61 cm). Using the centimeter side of the measuring tape, measure the head circumference by holding the tape from the subject's nasion to ~1 cm above the inion and then back to the nasion.
    NOTE: The nasion is the craniometric point between the eyes, whereas the inion is the tip of the external occipital protuberance.
    1. Select the correct net size that fits the subject's head circumference and soak it for at least 5 min (10 min maximum) into a liquid mixed solution composed of 1 qt of warm tap water, 1 tbsp of electrolytes (i.e., Potassium Chloride), and 1 tbsp of baby shampoo. During this soaking process, ensure that the net is turned inside out with the sponges facing out and the clasps fully loosened to fully immerse the sensors inside the plastic bucket containing the solution.
      NOTE: To ensure that the net's amplifier does not come near the solution and always remains dry, wrap a towel around the plug of the selected net and, if preferable, place it on a chair or support close to the sink where the plastic bucket is.
  4. Place five magnetic coils served as head position indicator (HPI) coils at known locations directly on the scalp of the subject using micropore paper tape: one on each side of the forehead near the hairline, one on each mastoid bone, and one on the top of the head.
    NOTE: The HPI coils define the position of the head relative to the superconducting quantum interference devices (SQUIDs) placed inside the MEG system by emitting known magnetic fields that can be localized during the scan. The number of HPI coils depends on the MEG system, but it usually ranges from 3-5 HPI coils.
  5. Place additional electrodes using tape to measure heart rate (electrocardiography, ECG), eye movements or blinks (electrooculography, EOG), and muscle activity (electromyography, EMG); the placement of these electrodes also allows the monitoring of the subject's health state.
    1. Place two ECG electrodes on the right and left side of the chest below the collarbones, respectively, to record the subject's heartbeat, and two EOG electrodes on the upper and lower side of the right eye, respectively, to record his/her vertical eye movements or blinks.
      1. To measure the muscle activity during the visuomotor task, additionally, wipe the subject's fingers with alcohol pads for better adhesion of the tape on the skin and tape a total of two pairs of non-disposable cup electrodes on each hand: one on the first dorsal interosseous (FDI) and one on the abductor pollicis brevis (APB).
      2. Before taping all these electrodes, place the conductive paste inside the electrode cup until it is slightly overfilled to reduce skin impedance and ensure an optimal blend of adhesiveness and conductivity.
    2. For tactile stimulation, attach thin elastic membranes directly to the distal, volar parts of three digits (i.e., thumb [D1], middle finger [D3], and pinky [D5]) of both hands. Inflate the membranes with compressed air pulses through rigid plastic tubes using an air puff stimulation device. Release the compressed air pulses with an inter-stimulus interval of 1.5 ± 0.5 s following a pseudorandom order. Adjust the pressure of the tactile stimulator to 50 psi.
  6. While the subject still sits on the wooden chair distant from any metallic object, determine the three-dimensional (3D) positions of the fiducial anatomical landmarks, five HPI coils, and other head shape points using a digitizer. During this head digitization process, ask the subject to sit comfortably, look straight ahead, and remain virtually motionless since small movements may affect the localization accuracy.
    1. Place the reference receiver through the plastic goggles (i.e., eyeglasses with the reference cube attached on one side) on the subject's head and adjust the clasps to ensure a fixed reference frame to the subject that must stay relatively still during the entire measurement.
    2. Through the primary stylus receiver, locate the fiducial anatomical landmarks (i.e., nasion and left/right preauricular points) and the HPI coils' position, and uniformly sample additional scalp points (at least 100, preferably close to ~500) to enhance a high-quality reconstruction of the head surface.
      NOTE: The fiducial anatomical landmarks define the subject coordinate system. The digitizer generates the coordinates of a sensor in the 3D space using one transmitter (typically mounted behind the subject on the back of the wooden chair) and two receivers (i.e., the stylus and reference receivers).
    3. After the digitization is completed, place the stylus receiver ~15 cm away from the subject and transmitter and digitize a random point to finish the digitization process.
      NOTE: This final step of the digitization process may differ from other commercial products.
  7. Before applying the EEG net, ask the subject to sit on a chair close to the EEG amplifier and place towels on his/her chest and shoulders to absorb eventual drips due to the net's application. Remove the EEG net from the plastic bucket, turn it with the sponges facing the inside, and gently wrap it around a towel to absorb the mixed solution in excess.
    1. With the subject sitting on the chair and instructed to keep his/her eyes closed during this step, put both hands inside the net and spread it using fingers, and finally place it on the subject's head.
    2. Without moving the positions of the HPI coils, adjust the stretched net on the subject's head using fingers to ensure that the reference and nasion channels are correctly positioned at the center of the scalp and between the subject's eyes, respectively, and finally fasten the chin strap once the net is in the correct position.
  8. Using an EEG impedance meter, ensure that all scalp-electrode impedances are in the 0-50 kΩ range (values ˂5 kΩ are recommended) to prevent any signal distortions. To reduce the scalp-electrode impedances, check that each electrode has good mechanical and electrical contact with the scalp by using either a wooden cotton swab to remove the subject's hairs between the electrode and the scalp or a disposable plastic pipette to transfer more conductive mixed solution into the sponges of the electrodes if needed.
    1. Once all the impedances are ideally up to 50 kΩ, disconnect the amplifier and prepare the subject for the EEG electrode digitization.
      NOTE: Perform the EEG electrode digitization outside the MSR and ensure enough room around the subject to manage the scanning process.
  9. Determine the 3D positions of the EEG electrodes using a handheld optical scanner. During this process, ask the subject to sit comfortably and look straight ahead unless instructed otherwise.
    1. Firstly, open the optical scanner software, select the sensor template that matches the EEG sensor layout used during the recordings, and then initiate the scanning process.
    2. During the scanning, hold the scanner at a certain distance from the EEG net (usually ~45 cm), with its scanning apertures perpendicular to the surface of the sensors, and slowly move it around the subject's head following arched swaths from the top (center of the head) to the bottom (last sensors row along the neck) to record the physical locations of all sensors.
      NOTE: The optical scanner digitizes the EEG electrode locations on the subject's head and transforms them as a 3D coordinate file; it is usually characterized by two optical sensors that emit infrared (IR) light sources. Each scanned location usually appears on a 3D sensor cloud. The 3D sensor cloud provides feedback for the scanning, probing, and alignment of the sensor positions, whereas the 2D sensor map provides feedback for the labeling of these sensor positions. The scanning process of the EEG electrode positions requires a total of 5-10 min, including the probing of the fiducial points. However, the scanning time may sometimes depend on the performance of the optical scanner in detecting the electrode positions.
  10. Once all the EEG electrodes have been scanned (at least 95%), probe the fiducial points (i.e., nasion and left/right preauricular points) and four alignment sensors (i.e., front, left, and right align nodes, and REF node) using the wireless optical probe to align the 3D sensors cloud to the selected sensor template.
    NOTE: The alignment sensors are numbered based on the EEG sensor net configuration.
    1. To probe the fiducial points, place the optical probe's tip on the subject's skin at the center of the fiducial point of interest by ensuring that the scanner's scanning apertures point at the probe's reflective disks. Similarly, place the optical probe's tip at the center of the alignment sensor of interest to probe the alignment sensors.
  11. Once all the sensors have been scanned and probed, review their positions and labels on the 3D sensor cloud and 2D sensor map with respect to the actual EEG net to check and eventually correct possible errors; if no errors occurred during the scanning process, export the 3D coordinate .txt file and convert it into the preferable format.
    NOTE: The 3D electrode coordinates are usually stored in .txt format and can be converted via optical scanner software in several formats (e.g., .xml, .sfp, .elp, or .nsi).
  12. After the EEG electrode digitization process (steps 2.9-2.11) is completed, prepare the subject to be transferred inside the MSR for performing resting/sleeping data (step 2.13), visuomotor task (step 2.14), auditory stimulation (step 2.15), and somatosensory stimulation (step 2.16).
  13. For the resting/sleeping data, set the gantry of the MEG system to the supine position (Figure 1A) and arrange the nonmagnetic and compatible bed so that the removable headrest is aligned with the helmet-shaped opening at the bottom of the dewar. After adjusting the bed to the correct position, set the brake valve of the bed to the off locked position. Place a sheet or blanket over the bed and a small foam pad pillow on the removable headrest for head fixation and comfort during the recording.
    NOTE: The dewar is a cryogenic storage container filled with liquid helium in which sensor arrays are spatially arranged at the bottom via a helmet-shaped opening designed to surround the subject's head. The helmet fits up to 59-61 cm head circumference. The gantry is the mechanical system supporting the dewar that allows modifying its elevation and angle based on the measurement position (i.e., seated or supine).
    1. Transfer the subject inside the MSR and help him/her sit on the edge of the bed and lie down on it. Place several blankets on the subject's body to keep him/her warm during the recording by ensuring that the electrodes' cables are easily accessible, and lightly fasten the safety belts (or pull up the railings if present), explaining to the subject that this step is to prevent him/her from rolling out of the bed while sleeping. If needed, place an additional rolled towel under the neck to provide support for the subject's neck and shoulders.
    2. Unlock the brake valve of the bed to gently move the subject's head, which is placed on the removable headrest under the helmet-shaped opening of the dewar until it touches the inside of the helmet. To increase the signal-to-noise ratio (SNR), bring the subject's head as close to the helmet as possible.
    3. Plug the HPI coils, ECG, and EOG electrodes on the corresponding panels of the MEG system, connect the EEG net to the amplifier unit inside the MSR, and check the measurements of the head's coordinates from the acquisition workstation outside the MSR to assess whether the subject's head is properly positioned under the dewar.
    4. With the subject's consent, reduce the intensity of the light inside the MSR to help stimulate relaxation and sleep. When the subject feels relaxed and comfortable, instruct the subject to rest with his/her eyes closed or sleep during the recording. Reassure the subject that he/she will be observed on the monitor outside the MSR via the radiofrequency-shielded color camera mounted on the wall of the MSR for the entire recording.
  14. For the visuomotor task, set the gantry of the MEG system to the upright position (Figure 1B) and arrange the MEG chair so that the subject's head is under the gantry, close to the helmet-shaped opening at the bottom of the dewar. After adjusting the chair to the correct position, set the brake valve of the chair to the off-locked position ("0").
    1. Transfer the subject inside the MSR. Help him/her to sit on the nonmagnetic and compatible chair and find a comfortable and relaxed position. Place several blankets on the subject's body to keep him/her warm during the recording by ensuring that the electrodes' cables are easily accessible, and place the removable table so that the subject can put his/her hands on it during the task. If needed, place a towel under the subject's knees to help maintain the seated position, and do not slide down.
      NOTE: Since the subject may relax during the visuomotor task and, therefore, assume a lower position than the initial one, carefully lift the chair at the end of each task session via the elevation pedal (if present) or place towels or blankets on the chair so that the subject's head touches the inside of the helmet again. If needed, place additional towels or blankets behind the subject's head not only for better comfort but also to help the subject keep the head as straight as possible. Visuomotor stimulation can be performed alternatively in a supine position to avoid moving the dewar in the middle of a recording session.
    2. Once the subject is in the right position, plug the HPI coils, ECG, EOG, FDI, and APB electrodes on the right panel of the MEG machine, connect the EEG net to the amplifier unit inside the MSR, and raise the chair through the elevation pedal (if present) with small movements or place additional towels or blankets on the chair until the subject's head lightly touches the inside of the helmet (check the measurements of the head's coordinates from the acquisition workstation outside the MSR).
    3. Place the projection screen, in which the visual stimuli will be projected via a projector mirror system positioned outside the MSR, in front of the subject (Figure 1B), and explain the visuomotor task to perform during the recording. Particularly, instruct the subject to tap his/her index finger on the table only when the visual stimulus (e.g., an image) appears on the screen, respectively, for the right and left hand. Make sure that the subject understands the task or feels comfortable performing it alone; if needed, practice the task together with the subject several times to help him/her get familiar with it.
      NOTE: If a visuomotor stimulation session is performed in a supine position, a mirror is placed at a distance above the subject's face to reflect the visual stimuli from the projector.
    4. Before closing the door of the MSR, ask the subject if he/she feels comfortable being alone inside the room; in case he/she does not, either one person from the team or his/her parents will remain inside the MSR during the recording sessions. Moreover, reassure the subject that he/she will be observed on the monitor outside the MSR for the entire recording.
  15. For the auditory stimulation, use the setup described in step 2.14 with the projection screen in front of the seated subject. Help the subject to wear the headphones (or earphones) through which the sound triggers (e.g., modulated chirp sounds) are delivered.
    1. Instruct the subject to fixate the stimuli (e.g., green point on a black background) projected on the screen while listening to the sound triggers. If needed, perform a training session to help the subject better understand the procedure. Before closing the door of the MSR, repeat the safety procedures as previously described.
  16. For the somatosensory stimulation, use the setup described in step 2.14. Ask the subject what video (or movie) he/she wants to watch on the projector screen in front of him/her.
    1. Instruct the subject to relax with eyes open, watch the selected video, stay as still as possible, and ignore the tactile stimuli delivered to his fingers during the recording. Explain to the subject that he/she will feel gentle taps on the skin at the tip of the fingers, respectively, for each hand. If the subject feels uncomfortable, perform a training session to reassure him/her.
      NOTE: Eye fixation on a visual target is a well-established technique to minimize biological artifacts that can affect the quality of the recording and distract the subject from the tactile stimuli delivered during the data acquisition.

3. Data acquisition

NOTE: The acquisition of simultaneous MEG and EEG data is performed in the MEG facility at Cook Children's Medical Center (CCMC). More details about the clinical use of MEG on pediatric children with epilepsy can be found elsewhere8,27,45.

  1. Record MEG signals using the whole-head MEG system (sensor coverage: 1,220 cm2) characterized by 306 channels grouped in 102 identical triple-sensor elements with one magnetometer and two orthogonal planar gradiometers. Set a sampling frequency of at least 1 kHz.
    NOTE: Single-coil magnetometers measure the component of the magnetic field perpendicular to the MEG helmet surface. Planar gradiometers consist of a "figure-of-eight" coil configuration characterized by pairs of magnetometers placed at a small distance between each other and measure the difference in the magnetic field between their locations (i.e., the difference between the two loops of "eight"), also named spatial gradient. Compared to magnetometers, planar gradiometers are less sensitive to deep brain sources but more robust in detecting superficial sources by suppressing environmental noise. These 306 channels are immersed and cooled down in liquid helium at -296 °C (4.2 K) to become superconductors.
  2. Record EEG signals simultaneously using the nonmagnetic 256-channel EEG net with Ag/AgCl electrode sensors evenly spaced over the scalp, cheeks, and back of the neck. Set a sampling frequency of at least 1 kHz.
  3. Close the door of the MSR to begin recording. Through the voice intercom system, communicate with the subject, checking if he/she feels comfortable being alone inside the MSR. Constantly monitor the subject on video and, in an emergency, immediately enter the MSR.
    NOTE: In case he/she does not feel comfortable or the massive MSR door is intimidating him/her, either one person from the team or his/her parents can remain inside the MSR during the recording sessions seated on a wooden chair close to the subject; ensure that all metallic objects are removed before entering the room.
  4. Before each recording, instruct the subject through the voice intercom system to keep his/her position still for ~30 s before starting the task. For the visuomotor task, additionally communicate via the intercom which index finger (either right or left) he/she will use for the first recording session.
    1. During this period in which the subject stays still, press the measure button from the HPI measurement dialog on the MEG data-acquisition system to measure the magnetic field generated by current fed into the HPI coils and determine the head measurement positions with respect to the MEG sensor array; therefore, confirm that the subject is well positioned (head |z-coordinate| ˂ 75 mm) and write down the 3D measurements for each session.
    2. If there is still space between the subject's head and the helmet, either re-enter inside the MSR and adjust the height of the chair via the elevation pedal (if present), place towels or blankets on the chair, or instruct the subject via the intercom on how to shift the head in a closer position to the helmet (if the subject is seated), and finally check the head measurement positions again. In case of resting/sleeping recordings, re-enter the MSR and bring the bed closer to the helmet, reducing the head-helmet spatial gap.
  5. Once the subject is well positioned with respect to the MEG helmet and ready to start, begin the first recording session (duration ~10 min) by following a precise order (see step 3.5.1) for accurate MEG and EEG synchronization (see step 3.12).
    NOTE: To ensure high-quality recordings, the first recording session is critical to capture artifacts present in the data due to movements of the subject or caused by the external environment. If needed, re-enter inside the MSR to adjust eventual channel connections or the subject's position on the MEG chair. It is recommended to take notes of any unusual artifacts or events during the recordings that can be reviewed afterward if needed.
    1. Press the recording button on the EEG data-acquisition software to begin the EEG recording. Press the recording button on the MEG data-acquisition software to begin the MEG recording. Finally, press the start button from the software of the stimulation computer to either display visual stimuli or deliver auditory stimuli.
      NOTE: The stimulation computer running the visual (or auditory) stimulus software is connected to the projector mirror system outside the MSR, which can be turned on or off depending on the type of recording performed. During resting/sleeping recordings, the system is turned off since the subject is either resting or sleeping, but running the visual stimulus software on the stimulation computer helps timing each recording session. During the visuomotor task, as well as during the auditory and somatosensory stimulations, the system is turned on by allowing the subject to observe the stimuli or a video projected on the screen placed in front of him/her while the software is running. For this study, it was selected a total of (i) 107 stimuli (i.e., 85 images overlaid on a checkerboard background and 22 checkerboard backgrounds) with a ~4 s interval between each stimulus for the visuomotor task; (ii) 200 modulated chirp sounds with an inter-stimulus interval of 3 s for the auditory stimulation; and (iii) 200 tactile stimuli for each finger (i.e., D1, D3, and D5) following a semi-random sequence with an inter-stimulus interval of ~1.5 s for the somatosensory stimulation.
  6. To stop the recordings, press the stop button on the MEG data-acquisition software and then the stop button on the EEG data-acquisition software. At the end of each recording session, communicate with the subject via the intercom to reassure him/her and if no channel connections or position adjustments inside the MSR are needed, proceed with the next session.
    1. For the visuomotor task, select different visual stimuli for each session to keep the subject motivated and entertained during the recording.
  7. For either visuomotor or resting/sleeping data, record a total of ~1 h simultaneous MEG and EEG recording, characterized by 5-6 sessions. However, the number of sessions may vary for each subject. Further, record a total of ~20 min (1-2 sessions of ~10 min each) and ~14 min (1-2 sessions of ~7 min each) of simultaneous MEG and EEG recording for the auditory and somatosensory stimulation data, respectively.
    NOTE: In this study, the MEG and EEG data are automatically stored at the end of the recordings as .fif and .mff formats, respectively, in the IT storage system of CCMC.
  8. When the recording ends, enter the MSR to help the subject stand up from either the chair or bed and ask him/her to sit on a chair outside the MSR to remove both the EEG net and electrodes.
    1. Instruct the subject to close his/her eyes until told otherwise and help him/her to remove the EEG net by fully loosening the chin straps and gently pulling out the net using two hands (from the forehead towards the back of the subject's head) until it is completely peeled away. During this step, ensure not to pull the subject's hair while removing the net. Additionally, help the subject to gently remove the remaining electrodes (i.e., ECG, EOG, and EMG in case of visuomotor task) previously taped on his/her skin.
  9. After the EEG net and electrodes are removed, inform the subject (and his/her parents) that all the procedures are finally completed.
  10. After the subject leaves the room, carefully clean the optical scanner (as described in the user manual) and store it inside its protective case.
    1. Clean and disinfect the surfaces of any equipment used during the recordings (e.g., chairs, beds, desks) with either hospital-approved hydrogen peroxide wipes or disinfecting spray and paper towels, put the used blankets and towels inside the container provided by the hospital, and throw away any piece of used tape.
    2. Store the measurement tools inside the storage cabinet and clean the inside of the electrode cups filled with conductive paste using wooden cotton swabs under running tap water.
    3. To rinse the EEG net, fill the plastic bucket in the sink with clean, warm tap water and repeat the following steps for a total of four times.
      1. Submerge the EEG net in the water and gently agitate the EEG net for 10-20 s (or dip it inside and outside the bucket ~25 times).
      2. Drain the water from the bucket, and refill the bucket with clean and warm tap water.
    4. To disinfect the EEG net, fill the plastic bucket in the sink with the disinfecting solution composed of 2 qt of lukewarm tap water and 1 tbsp of disinfectant and soak the EEG net inside it for 10 min. Rinse the bucket from the disinfecting solution and follow the rinse-and-drain process three times to remove any residual solution from the EEG net.
    5. For either the rinsing or disinfecting process, remove the immersed EEG net from the bucket, dry it by removing the excess water with the help of a clean, dry towel, and store it by hanging it close to the sink.
  11. To suppress internal and external magnetic interferences and measurement/movement artifacts from the MEG data, apply the temporal extension of the signal space separation (tSSS) method to the raw-data MEG .fif file.
    NOTE: The spatiotemporal Maxwell filtering (tSSS) is ideal for suppressing sources of interference located inside or very close to the MEG sensor array, namely internal interference.
  12. When performing simultaneous MEG and EEG recordings, spatially align the coordinate systems of the two acquisition devices with respect to the anatomical landmarks on the subject's head (Figure 2A) and correct the linear clock drift between signals that occurs due to the possible different sampling rates (Figure 2B).
    NOTE: During the recordings, both MEG and EEG signals can be also affected by slow shifts over time due to possible delays in pressing the Start and End buttons and to an internal clock drift that occurs when the triggers are sent to the MEG and EEG data-acquisition software. To ensure precise synchronization between these signals, an in-house code was developed in Python that uses the trigger events sent on both systems during the data acquisition as common trigger signal. The code includes three functions available in the MNE-Python software library: two functions that read the MEG and EEG signals and one function that extracts information of the trigger events from the signals, such as channel names and timestamps (i.e., date and time of the event occurrence). Differences in time between the occurrence of trigger events in each signal (i.e., delta) define the linear clock drift over time (Figure 2B). A detailed description of the developed code is provided in the next steps (see 3.12.1-3.12.4).
    1. Use the difference between the first trigger event occurring on each signal as offset value (i.e., portion to be trimmed from one of the two signals) to align the recordings.
      NOTE: The mne.io.read_raw_fif and mne.io.read_raw_egi functions convert the MEG and EEG recordings in a 2D array format, whereas the mne.find_events function extracts event information from the raw signals.
    2. Once these first triggers are aligned, compute the Pearson's correlation coefficient to assess the degree of correlation between the signals; p-values < 1 x 10-6 are recommended to ensure a perfect alignment.
      NOTE: The pearsonr function from the scipy library estimates the Pearson's correlation coefficient between MEG and EEG signals and the p-value of this correlation.
    3. To validate this correlation accuracy, estimate the drift rate between the two signals by performing a first-degree polynomial fit and use the resulted extent of disparity represented by the coefficient of the polynomial function to resample the signals on the x-axis of the polynomial fitting (Figure 2B).
      NOTE: The polyfit function from the numpy library fits the MEG and EEG signals within a polynomial function; this function returns a coefficient that represents the extent of disparity between the two signals. The mne.resample function resamples the MEG and EEG signals according to the coefficient of the polynomial function.
    4. After the resampling is complete, compare the timestamps of the last trigger event occurring on each signal and remove those time windows that are not common in both MEG and EEG signals. Finally, merge the synchronized MEG and EEG signals to create a single recording characterized by both MEG and EEG sensors that can be used for further analysis.
      NOTE: The mne.add_channels function merges the two signals to create a single recording.
  13. At the end of each ~1.5 h recording, use a total of 5-6 (~10 min each), 1-2 (~10 min each), and 1-2 (~7 min each) sessions of synchronized MEG and EEG recordings, respectively for the visuomotor (and resting/sleeping), auditory, and somatosensory stimulation data, for the data analysis.
    NOTE: Ideally, the subject needs to perform right-finger tapping for three visuomotor recordings and left-finger tapping for the remaining three visuomotor recordings.

4. Data analysis

  1. Mapping of the irritative zone
    1. Generate the 3D cortical surfaces from the MRI of the subject using the cortical reconstruction process of FreeSurfer, which is an open-source neuroimaging tool for processing, analyzing, and visualizing human brain MR images46.
    2. Import the reconstructed anatomy on Brainstorm, which is an open-source application of MATLAB dedicated to MEG and EEG data visualization and processing47, to visualize the cortical reconstruction results. From Brainstorm, set the fiducial points (i.e., nasion, left/right preauricular, anterior/posterior commissure, and interhemispheric) on the imported MRI that define the subject coordinate system.
    3. Import the simultaneous MEG and HD-EEG signal on Brainstorm and register the MEG and EEG sensors on the MRI using the MRI registration process to adjust their alignment to the digitized fiducial points. If necessary, project the EEG sensors on the cortical surface.
    4. Open the simultaneous MEG and HD-EEG recording and visually inspect the raw data to remove bad channels. Additionally, apply the signal-space projection (SSP) artifact-correction technique available in Brainstorm to reject biological artifacts (e.g., heartbeats, eye blinks) from the recordings.
    5. Apply notch (50 or 60 Hz, according to the power-line interference) and band-pass (1-70 Hz) filters to the simultaneous MEG and HD-EEG data.
    6. Select portions of data containing interictal activity characterized by frequent IEDs, such as spikes and sharp waves, and having minimal motion artifacts (if possible).
      NOTE: IEDs are transient waveforms characterized by a <70 ms temporal evolution, >50 µV amplitude, and a sharp shape that can be clearly distinguished from the background activity in the 1-70 Hz frequency band48. Figure 3A represents an example of portions from the simultaneous MEG and HD-EEG signals with frequent IEDs visible on both recordings.
    7. Using the standard display setting of 10 s per page, mark the negative peak of each IED that occurs on both MEG and EEG recordings (Figure 3A), as well as on each modality alone. Before marking each IED, check the topography field and potential maps for the MEG and EEG, respectively.
      NOTE: Voltage distributions of additional time points during the rising phase of each IED, rather than its peak, must also be checked to account for possible propagating epileptic activity49. More details on how to mark interictal spikes events using Brainstorm can be found elsewhere (https://neuroimage.usc.edu/brainstorm/Tutorials/Epilepsy).
    8. Compute a realistic head model, defined as a three-layer (i.e., scalp, inner skull, and outer skull) geometric model (Figure 3B), using the OpenMEEG BEM (symmetric boundary element method) software available in Brainstorm. Use the MRI volume as source space (source-point grid with 5 mm spatial resolution).
      NOTE: The OpenMEEG BEM software uses the symmetric boundary element method to compute a realistic forward model characterized by the scalp (i.e., air-scalp interface), outer skull (i.e., scalp-skull interface), and inner skull (skull-brain interface). As an alternative solution, the finite element method (FEM) can be used to solve the forward problem since it allows a realistic subject-specific representation of the head volume conductor. Head tissue conductivity values are often assumed from the literature and may vary depending on the age of the subject50. To solve the forward model using the FEM, compute the realistic head model, defined as a three- up to five-layer (i.e., white matter, gray matter, cerebrospinal fluid, skull, and skin) geometric model, using the DUNEuro FEM software available in Brainstorm47,50,51. More details on the estimation of the forward model using FEM can be found elsewhere (https://neuroimage.usc.edu/brainstorm/Tutorials/Duneuro). 
    9. Localize the underlying generators of the selected interictal spikes using the unconstrained ECD method on the MEG, EEG, and combined MEG and EEG sensors array separately. To localize these dipole sources, compute the noise covariance from the empty-room MEG recordings or set it as an identity matrix.
      NOTE: Consider performing source localization on averaged IEDs with similar voltage distributions as an alternative approach in case the SNR of IEDs is low12.
    10. Perform the dipole scanning method available in Brainstorm at the peak of each IED previously marked to select the most significant source dipoles across the entire brain volume. Select only source dipoles with a goodness of fit >60% and estimate their clusterness defined for each dipole as the number of dipoles located at 15 mm from its center (Figure 3C).
      NOTE: More details about the ECD clusterness can be found elsewhere52.
  2. Mapping of the SOZ
    1. Generate the 3D cortical surfaces from the MRI of the subject using the cortical reconstruction process of FreeSurfer46.
    2. Import the reconstructed anatomy on Brainstorm47 to visualize the cortical reconstruction results. Set the fiducial points on the imported MRI that define the subject coordinate system.
    3. Import the simultaneous MEG and HD-EEG signal (containing the ictal event) on Brainstorm and register the MEG and EEG sensors on the MRI using the MRI registration process to adjust their alignment to the digitized fiducial points. If necessary, project the EEG sensors on the cortical surface.
      NOTE: If the subject had a seizure during the recordings, review the notes taken during the recording to assess in which session the ictal event was recorded.
    4. Open the simultaneous MEG and HD-EEG recording and visually inspect the raw data to remove bad channels. Apply the SSP artifact-correction technique available in Brainstorm47 to reject biological artifacts (e.g., heartbeats, eye blinks) from the recording.
    5. Apply notch (either 50 or 60 Hz, according to the power-line interference) and band-pass (1-70 Hz) filters to the simultaneous MEG and HD-EEG data.
    6. According to the recording notes, identify the onset of the seizure and its termination (if noted) and mark these events on the signal to identify the correct portion of data containing ictal epileptiform discharges. Examples of a seizure onset are displayed in Figure 4 for MEG and HD-EEG, respectively.
      NOTE: When a subject has a seizure during the MEG acquisition, medical personnel are required to immediately assist the subject and provide emergency care. Thus, the duration of those data portions with ictal events may be short.
    7. Using the standard display setting of 10 s per page, mark the negative peak of each burst of epileptiform discharges (e.g., repetitive spikes, sharp waves, or spike-wave complexes) occurring during the ictal event on MEG and EEG, as well as on each modality alone. Before each peak marking, check the topography field and potential maps for the MEG and EEG, respectively (Figure 4A).
      NOTE: Seizures are classified as generalized or focal depending on where their onset starts. Although dependent on the epilepsy syndrome, a seizure is a phenomenon of at least 10 s characterized by repetitive electrical discharges that may vary in frequency, amplitude, and morphology.
    8. Compute the three-layer (i.e., scalp, inner skull, and outer skull) realistic head model for the MEG and EEG sensors using the OpenMEEG BEM software available in Brainstorm. Use the MRI volume as source space (source-point grid with 5 mm spatial resolution).
      NOTE: To solve the forward model using the FEM, compute the realistic head model, defined as a three- up to five-layer (i.e., white matter, gray matter, cerebrospinal fluid, skull, and skin) geometric model, using the DUNEuro FEM software available in Brainstorm47,50,51. More details on the estimation of the forward model using FEM can be found elsewhere (https://neuroimage.usc.edu/brainstorm/Tutorials/Duneuro). 
    9. Localize the underlying generators of the selected ictal discharges using the unconstrained ECD method on the MEG, EEG, and combined MEG and EEG sensors array separately. To localize these dipole sources, compute the noise covariance from the empty-room MEG recordings or set it as an identity matrix.
    10. Perform the dipole scanning method available in Brainstorm at the peak of each ictal waveform previously marked to select the most significant source dipoles across the entire brain volume. Select only source dipoles with a goodness of fit >60% and estimate their clusterness (distance of dipoles from each dipole's center set to 15 mm)52.
  3. Mapping of the eloquent cortex
    1. Perform the previously described steps 4.2.1-4.2.3 using the signals recorded during the visuomotor task, as well as during the auditory and somatosensory stimulations.
    2. Open the simultaneous MEG and HD-EEG recordings and visually inspect the raw data to remove bad channels. Separately, for each session, apply the SSP artifact-correction technique available in Brainstorm47 to reject biological artifacts (e.g., heartbeats, eye blinks) from the recordings. Additionally, discard or mark as "Bad interval" all those data segments contaminated by artifacts that cannot be used for further analysis.
    3. Apply notch (either 50 or 60 Hz, according to the power-line interference) and band-pass (1-100 Hz) filters to the simultaneous MEG and HD-EEG data.
    4. For mapping the motor cortex, open the EMG signals recorded from the FDI and APB electrodes and manually mark the tapping event performed by the subject, separately for the right and left hand, by selecting the first peak of muscle activation different from the baseline on the FDI pair-electrode. Perform this step separately for each session using Brainstorm47.
      NOTE: Use the Filter tab for visualization only on Brainstorm and select frequency ranges (e.g., high-pass: 30 Hz; low-pass: 300 Hz; notch: 60 Hz) that can help to correctly mark the peaks of the EMG activity while tapping. The name of the events needs to be the same across sessions; for example, events when the subject performed the right-finger (or left-finger) tapping can be named as "Tap_right" (or "Tap_left"). This step allows us to perform a unique analysis on the same type of event that occurred across all sessions.
    5. From each session, import the events occurring on artifact-free segments by selecting the following parameters: visual stimuli: [-200; +500] ms time window, including the baseline of [-200; 0] ms before the stimulus onset; tapping events: [-1500; +1000] ms time window, including the baseline of [-1500; -1000] ms before the visual stimulus onset; modulated chirp sounds: [-500; +1000] ms time window, including the baseline of [-500; 0] ms before the stimulus onset; and tactile stimuli: [-100; +500] ms time window, including the baseline of [-100; 0] ms before the stimulus onset.
    6. Once the events of a specific task have been imported from all sessions, apply the average reference montage to increase the SNR on the EEG data and estimate the average across stimuli to obtain the event-evoked fields and potentials. In the case of motor tasks (i.e., tapping) and tactile stimulation, this last step is performed for the right and left hand, respectively.
      ​NOTE: Panels A and B of Figure 5, Figure 6, Figure 7, and Figure 8 show examples of visual-, motor-, auditory-, and somatosensory-evoked responses, respectively, for MEG and EEG and their relative topography field and potential maps. The total number of stimuli for a specific task strictly depends on the number of completed sessions; in the case of motor task, this number also depends on the correctly performed tapping task.
    7. For either the visual-, motor-, auditory-, or somatosensory-evoked fields and potentials, compute the three-layer (i.e., scalp, inner skull, and outer skull) realistic head model for the MEG and EEG sensors using the OpenMEEG BEM software available in Brainstorm. Use the 3D cortical surface as source space.
      NOTE: To solve the forward model using the FEM, compute the realistic head model, defined as a three- up to five-layer (i.e., white matter, gray matter, cerebrospinal fluid, skull, and skin) geometric model, using the DUNEuro FEM software available in Brainstorm47,50,51. More details on the estimation of the forward model using FEM can be found elsewhere (https://neuroimage.usc.edu/brainstorm/Tutorials/Duneuro). 
    8. For each event-evoked fields and potentials, compute the cortical sources on the averaged events using dSPM implemented in the Brainstorm toolbox for the MEG, EEG, and combined MEG and EEG sensors array separately. Set the noise covariance matrix using the pre-stimulus baseline period specific for each task (see step 4.3.5).
    9. Depending on the task, extract the maximum cortical sources observed at the eloquent brain area involved [primary visual cortex (V1), primary motor cortex (M1), primary auditory cortex (A1), or primary somatosensory cortex (S1)] that can be defined either using an atlas or through the reconstruction of a virtual sensor (i.e., region of interest, ROI) in that location (Figure 5C, Figure 6C, Figure 7C, and Figure 8C).
    10. Perform the Morlet wavelet time-frequency decomposition on the event-evoked field and potential sources at the eloquent brain area involved (V1, M1, A1, or S1) with a linear scale (frequency range: 1:1:100 Hz).
      NOTE: To estimate evoked oscillations that are time- and phase-locked to the stimulus onset, the MEG/EEG signal is initially averaged over events and then becomes subject to time-frequency analysis53,54.
    11. Perform standardization of the cortical source maps using the event-related perturbation method available in Brainstorm to normalize them with respect to the baseline (see step 4.3.5), separately for each event (Figure 5D, Figure 6D, Figure 7D, and Figure 8D).
      NOTE: Standardization of source maps using event-related perturbations is a normalization measure for time-frequency power maps that estimates the deviation from the mean over the baseline in percent. More details of this standardization process can be found elsewhere (https://neuroimage.usc.edu/brainstorm/Tutorials/TimeFrequency#Normalized_time-frequency_maps). 
  4. Validation
    1. Acquire extra-operative iEEG recording through the implantation of subdural grids and/or depth electrodes. The number, type, and placement of intracranial electrodes are decided by the multidisciplinary epilepsy team based on the clinical hypothesis generated from the summation of information obtained from the noninvasive diagnostic tests of the presurgical evaluation.
      NOTE: As part of the presurgical evaluation, extra-operative iEEG is recorded across several days via a digital EEG system using a 1,024 Hz sampling rate.
    2. Identify the SOZ as defined by an expert epileptologist based on the clinical information available for each subject.
      NOTE: The SOZ is defined as the brain area showing the first unequivocal ictal electrographic discharge, which is different from the background activity that can occur either prior to or concomitant with the clinically defined ictal onset. All channels involved in this ictal electrographic discharge are regarded as SOZ contacts, even if capturing seizures that originate from different brain areas.
    3. During the ictal event recorded on iEEG, mark the peak of each burst of epileptiform discharges occurring at each intracranial electrode and perform source localization on these ictal events using the ECD method as previously described in steps 4.1.7-4.1.10.
    4. Compare the iEEG localization results with respect to the clinically defined SOZ as the gold standard for the source localization results.

结果

$$\rightleftharpoonup{xx}$$ $$\longleftharp{xx}$$, $$\longrightharp{xx}$$,

来自库克儿童医疗系统（CCHCS）简和约翰·贾斯汀心理健康研究所癫痫门诊的药物难治性癫痫（DRE）儿科患者被纳入研究。本文展示了三位代表性患者的资料：（i）一名10岁女性，（ii）一名13岁男性，以及（iii）一名10岁女性。

病例1：一名10岁女性因自3岁起出现癫痫发作而入院。尽管已使用8种抗癫痫药物（ASMs），患者仍每日发作。初期发作表现为眼球偏斜（侧别不明确）和行为停止。随后，患者每日出现约持续30秒的癫痫发作，表现为发作性噘嘴（“警帽征”，chapeau de gendarme sign）、头向左侧偏转以及双侧上肢强直性僵硬（以右侧为主）。长程视频脑电图（EEG）显示两组成不对称的强直性发作，头向左偏，随后左上肢抬起。睡眠中亦观察到3次强直性发作，伴有频繁的全脑性快速多棘波和慢波发放，间歇性睁眼、上视以及左或右上肢抬高。这些多棘波和慢波睡眠期放电主要起源于左侧中颞叶。脑部MRI显示以下多灶性皮质发育不良：（i）左侧顶叶（中央后回）局灶性皮质发育不良（FCD），伴横贯纤维征（transmantle sign），属II型FCD；（ii）右侧顶枕交界区FCD；（iii）左侧颞极FCD。正电子发射断层扫描（PET）显示左侧顶叶、左侧颞叶及右侧顶枕交界区呈低代谢，与MRI上信号异常区域（即FCD病灶）相对应。患者被诊断为难治性癫痫，发作半侧综合征表现为典型的“警帽征”后继发上肢强直，提示可能起源于内侧额叶或岛叶/颞叶区域。建议进行广泛的双侧立体定向脑电图（sEEG）探查，靶区包括额叶、扣带回、岛叶及各发育不良区域。在颅内脑电图（iEEG）监测期间，患者出现典型发作，表现为“警帽征”后继发右侧或左侧上肢强直性抬高或屈曲，脑电图起始广泛，最大放电位于双侧前岛叶。多灶性发作间期放电（IEDs）主要见于双侧前颞叶及背外侧额叶皮层，包括双侧岛叶。基于iEEG记录所进行的电刺激成像（ESI）进一步确认了癫痫发作起始区（SOZ）的位置，临床定义为双侧背外侧额叶皮层及前岛叶。

作为术前评估的一部分，对同步采集的脑磁图（MEG）和高密度脑电图（HD-EEG）数据进行了源定位分析。MEG和HD-EEG记录显示双侧额颞区频繁出现癫痫样放电（IEDs）。图3A展示了MEG和HD-EEG数据中一次典型IED的示例；两种模态的地形电场图和电位分布图均提示右侧额颞区域可能存在潜在放电源。电生理源成像（ESI）结果显示偶极子呈弥散性分布，涉及双侧额颞区及顶叶区域。磁源成像（MSI）结果显示偶极子集中在右侧额颞叶，位置靠近右侧岛叶。联合电磁源成像（EMSI）结果显示双侧额颞区存在局灶性偶极子聚集，这与基于颅内脑电图（iEEG）金标准所进行的ESI结果一致，证实了临床观察结果（图3C）。通过EMSI估算的这些偶极子与iEEG定义的癫痫发作起始区（SOZ）之间的平均距离为9.81 mm（中位数：11.18；标准差：2.37）。

病例2：一名13岁男性因难治性癫痫入院，该患者自9岁起开始出现癫痫发作。发作起始表现为先兆，随后出现向左的头眼偏斜，有时意识保留，伴有左侧头部局灶性阵挛，持续约30秒，每周发作数次。曾尝试多种抗癫痫药物（ASMs），但均未能控制发作。长程视频脑电图（EEG）显示右侧后颞区棘波，以及右侧半球频繁的棘-慢波放电，累及中颞区、额颞区、颞顶区和中央顶区皮层。患者共记录到6次电-临床发作，表现为行为改变、头眼向左偏斜伴左上肢伸展，有时出现左上肢阵挛性活动；另有3次发作进展为双侧强直阵挛性发作。发作起始部位主要位于右侧中颞叶，并在右侧额颞叶区域出现演变。脑部MRI显示右侧大脑半球广泛皮层发育畸形（以岛盖区为主），伴有右侧大脑半球轻度容积缩小及右侧侧脑室代偿性扩张。患者被诊断为起源于右侧大脑半球的难治性癫痫，倾向于在弥漫性皮层发育不良区域内的颞叶及岛盖区起始。为明确病变累及范围，进行了立体定向脑电图（stereo-EEG）检查，电极植入右侧颞叶、岛盖区、岛叶及顶枕皮层。在颅内脑电图（iEEG）监测期间捕捉到多次电-临床局灶性起始发作，起始部位主要集中于右侧额颞叶较大范围区域。基于iEEG数据进行的癫痫源成像（ESI）将这些发作源定位在一个更为局限的区域，涉及右侧颞叶（靠近右侧中颞回）及岛盖区。

作为术前评估的一部分，在同步进行脑磁图（MEG）和高密度脑电图（HD-EEG）检查期间，患者共经历了两次癫痫发作：一次发生在数字化过程中坐在木椅上时，另一次发生在正式记录期间，其发作起始在MEG和HD-EEG上均可清晰观察到（图4A）。发作起始时刻的地形场和电位图显示，癫痫发作起始的潜在发生源可能位于右侧中颞叶，如图4A所示。对发作事件的源定位分析中，电生理源成像（ESI）与磁源成像（MSI）呈现出不同的结果：ESI显示偶极子主要集中于右侧额颞叶及中央顶叶区域，而MSI显示偶极子高度聚集，主要位于右侧颞叶（图4B），并在额颞皮层区域存在额外散在的偶极子。通过整合这两种方法的结果，联合源成像（EMSI）揭示出发作起始区定位于颞叶，该结果与颅内脑电图（iEEG）金标准下的ESI定位一致（图4B）。特别是，EMSI所得定位结果与iEEG监测所定义的癫痫发作起始区（SOZ）之间的平均距离为12.21 mm（中位数：13.62 mm；标准差：2.37 mm）。

病例3：一名15岁女性患者，患有局灶性特发性癫痫，因癫痫发作入院。该患者自13岁起出现癫痫发作，但回顾性分析提示可能早在8至9岁时即已发病，当时因反复、刻板的颈部运动被诊断为抽动症。患者表现为短暂的头部向左倾斜，有时进展为伴有过度运动行为的局灶性意识障碍性癫痫发作（即继发全面性强直-阵挛发作），并伴有夜间惊厥性发作。曾尝试多种抗癫痫药物（ASMs）治疗，但未能实现完全的发作控制。在长程视频脑电图（EEG）监测期间，患者出现局灶性电-临床发作，并继发全面性放电，发作起源于左侧后颞叶；同时记录到多次短暂的局灶性运动性发作，表现为头部向左倾斜；另有一次细微的脑电图发作，起始于左侧中央顶区皮层。脑部MRI未发现急性颅内异常，但存在Chiari I型畸形。头部正电子发射断层扫描-计算机断层扫描（PET-CT）检查结果为阴性。建议进一步检查，包括发作期单光子发射CT（SPECT）、同步脑磁图（MEG）与高密度脑电图（HD-EEG）、颈椎X线片、头颈部磁共振血管成像（MRA），以及最终对左侧大脑半球进行立体定向脑电图（sEEG）探查。

作为评估的一部分，患者接受了同步的脑磁图（MEG）和高密度脑电图（HD-EEG）记录，以定位重要的脑功能区，如初级视觉、运动、听觉和体感皮层。首先，患者完成了一项视觉运动任务，随后进行听觉和体感刺激。视觉刺激后，MEG 和 HD-EEG 均在约 70 毫秒（ms）时首次观察到皮层反应（图 5A）。图 5B 分别展示了 MEG 和 HD-EEG 在视觉刺激过程中所涉及皮层区域的场拓扑图和电位图。对于 HD-EEG，在覆盖枕叶脑区的通道中观察到极性反转；而在 MEG 中，相同区域表现出更复杂的场分布（图 5B）。采用 dSPM 进行源定位分析显示，在此时间点，Desikan-Killiany 图谱中的以下脑区出现局灶性皮层活动：（i）MSI 的楔叶；（ii）ESI 的外侧枕叶皮层；以及（iii）EMSI 的楔叶和外侧枕叶皮层（图 5C）。对视觉皮层反应进行时频分析发现，MSI（约 30–50 Hz）、ESI（约 40–50 Hz）和 EMSI（约 30–50 Hz）在伽马频段均出现事件相关同步化（ERS）（图 5D）。在运动诱发反应中，运动起始期间对侧 M1 区域的 μ 节律活动受到抑制（图 6A）。在 图 6B 中，我们分别展示了 MEG 和 HD-EEG 在运动任务期间激活脑区的场拓扑图和电位图。MEG 场图显示对侧中央脑区磁通入与磁通出发生明显变化，提示对侧 M1 区可能存在潜在的局灶性发生器（图 6B）。HD-EEG 电位图在相同区域显示出局灶性极性变化，且电位方向垂直于磁场（图 6B）。在执行敲击任务时，MSI、ESI 和 EMSI 的最大源激活峰值均出现在 Desikan-Killiany 图谱中对侧中央前回，如 图 6C 所示。在预期即将发生敲击动作期间出现的运动相关皮层反应中，MSI（约 20–30 Hz）和 EMSI（约 20–40 Hz）在 β 和 γ 频段出现 ERS，而 ESI 仅在 γ 频段（约 30–50 Hz）出现 ERS，文献中将此现象称为 μ 节律抑制（图 6D）55,56。听觉刺激诱发的 MEG 场和 HD-EEG 电位分别在刺激开始后约 80 毫秒和 120 毫秒出现最大正向峰值（图 7A）。在 图 7B 中，我们分别报告了 MEG 和 HD-EEG 在听觉刺激过程中所涉及皮层区域的场拓扑图和电位图。在 MEG 和 HD-EEG 中，覆盖左侧颞叶脑区的传感器均表现出明显的极性变化，具有清晰定义的负极和正极；这些相互垂直的磁场和电位图可能提示 V1 区存在潜在的局灶性发生器（图 7B）。对平均听觉诱发场和电位进行源定位分析发现，MSI、ESI 和 EMSI 的最大皮层激活均出现在 Desikan-Killiany 图谱中的横颞回及颞上回后部（图 7C）。对听觉诱发反应进行时频分析发现，MSI（约 40–60 Hz）和 EMSI（约 35–50 Hz）在 γ 频段出现 ERS，而 ESI 在 β 和 γ 频段（约 25–60 Hz）均出现 ERS（图 7D）。最后，MEG 和 HD-EEG 分别在触觉刺激开始后约 60 毫秒和 50 毫秒观察到首个皮层活动（图 8A）。在 图 8B 中，我们分别展示了 MEG 和 HD-EEG 在体感刺激过程中激活脑区的场拓扑图和电位图。MEG 场图显示，覆盖对侧顶叶区域的传感器处磁通量发生明显极性变化和显著改变；而 HD-EEG 电位图在相同区域的极性变化较不明显，且正极强度大于负极。这些相互垂直的磁场和电位图可能提示 S1 区存在一个局灶性皮层发生器。对平均体感诱发反应采用 dSPM 进行源分析发现，MSI、ESI 和 EMSI 在此时间点的最大皮层源活动均位于 Desikan-Killiany 图谱中的对侧中央后回（图 8C）。在触觉刺激下，MSI（约 15–40 Hz）和 EMSI（约 20–40 Hz）在 β 和 γ 频段出现 ERS，而 ESI 在 γ 频段（约 30–40 Hz）出现 ERS（图 8D）。

脑成像实验装置实景：受试者正在参与研究的脑磁图（MEG）系统。
图1：CCHCS实验室中MEG与高密度脑电图（HD-EEG）同步记录的实验装置。（A）高密度脑电图系统（256通道）与脑磁图系统（306个传感器），MEG设备的机架调整为仰卧位（90°，水平位置），以便使用无磁性、兼容MEG的病床记录静息或睡眠状态下的脑活动。技术人员正在为受试者（一名9岁女孩）准备记录过程，同时确保其安全与舒适。（B）使用无磁性、兼容MEG的座椅，将HD-EEG与MEG系统设置为坐姿记录模式。技术人员正在为受试者准备记录，并确保其在屏幕前的正确坐姿，视觉刺激将在视觉运动任务期间投射到屏幕上。请点击此处查看该图的放大版本。

MEG与高密度EEG系统中的共配准和线性漂移分析；示意图与图表。
图 2：使用不同采集系统整合同步MEG与高密度EEG记录数据的技术细节。（A）针对一名代表性受试者（一名9岁女孩），将MEG与高密度EEG传感器在相同坐标系（由受试者头部坐标定义）中进行空间对齐（共配准）。受试者的头部坐标由以下标志点表示：鼻根点（绿色）、左/右耳前点（分别为红色和蓝色）。图中显示了306个MEG传感器（蓝色）——包括102个磁强计和204个平面梯度计——以及头部位置指示（HPI）线圈（品红色）；同时，在同一坐标系中对齐的256通道高密度EEG也一并显示（粉红色）。（B）左侧面板：代表性受试者（一名9岁女孩）MEG与高密度EEG系统之间数据采样发生的线性漂移（即delta，以黑线表示）。Delta定义为同一触发信号发送至MEG与EEG系统的时间差的绝对值，其值随时间持续增加：从较低值（delta = 0 ms）到较高值（delta = 197 ms）。图中用蓝色虚线表示通过多项式函数估计并用于信号校正的线性漂移；校正后的漂移（随时间delta ≈ 0 ms）以红色虚线表示，代表MEG与EEG系统之间的时间同步。右侧面板：显示了对发送至MEG与EEG系统的最后一个触发信号估计的时间偏移（delta = 197 ms）的图形化表示。请点击此处查看该图的放大版本。

通过同步 MEG 和 EEG 进行痫样放电间期尖波分析；源定位，ECD 聚类图。
图 3：MEG 与高密度 EEG（HD-EEG）数据中的痫样放电间期（IEDs）。（A）一名10岁女性患者（病例1）同步记录的 MEG 与 HD-EEG 数据中的一段时程（10秒），其 IEDs 出现频繁。为便于可视化展示，从中选取了306个 MEG 传感器中的一部分以及256个 EEG 电极中的一部分。在子图中分别展示了 IED 峰值时刻 MEG 和 HD-EEG 的磁场拓扑图与电位分布图。（B）MEG 与 HD-EEG 传感器（黄色）在被试个体三维头部及皮层表面（蓝色）上的共配准位置。基于术前 MRI 构建的包含三层结构的真实边界元法（BEM）头部模型，三层分别为头皮（灰色）、外层颅骨（黄色）和内层颅骨（粉色）。（C）基于等效电流偶极子（ECD）方法对 IEDs 进行源定位聚类分析的结果，显示于被试个体的术前 MRI 上，包括 ESI、MSI、EMSI 以及基于 iEEG 的 ESI（金标准）52。偶极子聚类程度的热图显示拟合优度 >60% 的区域，颜色由低值（蓝色）到高值（红色）表示。通过 iEEG 数据进行 ESI 分析所确定的癫痫发作起始区被视为金标准（橙色和绿色圆圈）。请点击此处查看该图的放大版本。

MEG、高密度脑电图的癫痫发作起始分析图及源定位MRI、发作事件、脑区聚类。
图4：MEG与高密度脑电图（HD-EEG）数据中的癫痫发作起始。（A）一名13岁男性患者（病例2）的MEG与HD-EEG同步记录片段（10秒），显示癫痫发作起始时刻（红色箭头）。为便于可视化，从中选取了306个MEG传感器和256个EEG电极的一部分进行展示。发作起始时刻的MEG和HD-EEG拓扑场图与电位图分别以内嵌子图形式显示。（B）采用等效电流偶极子（ECD）方法在发作起始时刻进行源定位聚类分析的结果，显示于该受试者的术前MRI图像上，包括ESI、MSI、EMSI以及基于iEEG的ESI（金标准）52。偶极子聚类程度的热图显示拟合优度>60%的区域，颜色从低值（蓝色）到高值（红色）变化。通过iEEG数据进行ESI所定义的癫痫发作起始区被视为金标准（蓝色圆圈）。请点击此处查看该图的放大版本。

使用 MEG 和 HD-EEG 进行视觉诱发电位分析；激活图谱；时频分析结果。
图 5：来自 MEG 和 HD-EEG 数据的视觉诱发电场和电位。 （A）显示了一名 15 岁女性在 -100 ms 至 300 ms 时间区间内 MEG（上图）和 HD-EEG（下图）的平均视觉诱发电反应。（B）分别显示了初级视皮层的 MEG 场拓扑图和 HD-EEG 电位图。（C）使用动态统计参数映射（dSPM）方法对 MSI、ESI 和 EMSI 估计的 Desikan-Killiany 图谱脑区（即楔叶和外侧枕叶皮层）皮层激活最大振幅的源激活图。显示了源激活（dSPM 标准化 z 分数）的热图。（D）在 -100 ms 至 300 ms 时间窗内，对初级视皮层的视觉诱发电反应进行 Morlet 小波时频分解所得到的时频图。显示了时频功率的热图，其值以标准化数据相对于基线 [-200; 0] ms 均值的偏差百分比表示。请点击此处查看该图的放大版本。

运动诱发电位图、MEG-EEG拓扑图、脑激活图、时频分析。
图6：来自MEG和高密度脑电图（HD-EEG）数据的运动诱发电场和电位。 （A）显示了一名15岁女性在左手食指敲击任务中，时间区间为-100至300 ms的平均运动诱发电位响应，上图为MEG结果，下图为HD-EEG结果。中间面板显示了同一时间区间（-100 ms至300 ms）内的肌电图（EMG）信号及运动起始时刻（紫色箭头）；该信号经30–300 Hz频段滤波处理（陷波滤波：60 Hz）。（B）分别展示了MEG和HD-EEG对应的初级运动皮层的电场拓扑图和电位分布图。（C）利用动态统计参数映射（dSPM）方法估计出的MSI、ESI和EMSI在Desikan-Killiany图谱对侧中央前回处皮层激活最大振幅的源激活图。热图显示了源激活强度（dSPM标准化z分数），并叠加显示中央沟（黑线）。（D）采用Morlet小波时频分解方法，在初级运动皮层的运动诱发电位响应上获得的时间窗为-300 ms至500 ms的时频图。热图显示了时频功率变化，以标准化数据相对于基线期[-1500; -1000] ms均值的偏差百分比表示。请点击此处查看此图的放大版本。

听觉反应；MEG、高密度脑电图（HD-EEG）图谱；脑地形图、激活图；时频分析。
图7：来自MEG和高密度脑电图（HD-EEG）数据的听觉诱发电场和电位。（A）显示了一名15岁女性在-100 ms至300 ms时间区间内MEG（上图）和HD-EEG（下图）的平均听觉诱发电反应。（B）分别显示了MEG和HD-EEG对应的初级听觉皮层的脑地形图电场与电位图。（C）利用动态统计参数映射（dSPM）方法估计的MSI、ESI和EMSI在Desikan-Killiany图谱的颞横回及颞上回后部皮层激活最大振幅的源激活图。显示了源激活的热图（dSPM标准化z分数）。（D）在-100至300 ms时间窗内，对初级听觉皮层的听觉诱发电反应进行Morlet小波时频分解所得的时频图。显示了时频功率的热图，其值以标准化数据相对于基线[-500; 0] ms均值的偏差百分比表示。请点击此处查看该图的放大版本。

体感诱发电反应、MEG 和高密度 EEG 数据、地形图、激活、时频分析。
图 8：来自 MEG 和高密度 EEG 数据的体感诱发电场和电位。 （A）显示了一名15岁女性在左手手指刺激条件下，MEG（上图）和高密度 EEG（下图）在 -100 至 300 ms 时间区间内的平均体感诱发电反应。（B）分别显示了 MEG 和高密度 EEG 的初级体感皮层的电场和电位地形图。（C）利用动态统计参数映射（dSPM）方法估计的 MSI、ESI 和 EMSI 在 Desikan-Killiany 图谱对侧中央后回处皮层激活最大振幅的源激活图。显示了源激活（dSPM 标准化 z 分数）的热图，并叠加中央沟（黑线）。（D）在初级体感皮层的体感诱发电反应上，采用 Morlet 小波时频分解方法获得的 -100 ms 至 300 ms 时间窗内的时频图。显示了以基线 [-100; 0] ms 期间标准化数据相对于均值的偏离百分比表示的时频功率热图。请点击此处查看该图的放大版本。

讨论

$$\rightleftharpoonup{xx}$$ $$\longleftharp{xx}$$, $$\longrightharp{xx}$$,

本研究展示了在儿童癫痫持续状态（DRE）患者静息/睡眠、执行任务或接受刺激时，同步记录脑磁图（MEG）和高密度脑电图（HD-EEG）的实验设置，并提出了一种利用联合源成像（EMSI）以及个体化的MEG源成像（MSI）和EEG源成像（ESI）来定位激惹区、癫痫发作起始区（SOZ）及功能重要脑区的方法学框架。我们进一步提供了技术建议，用于整合来自不同商用设备的MEG与HD-EEG数据，这些设备各自具有独特的特性。我们展示了三个病例的数据，以增强EMSI在定位致痫区和功能重要脑区方面的临床实用性。研究结果表明，EMSI的效果优于单独使用任一模态的结果，这很可能归因于MEG与EEG信号在联合分析中互补特性的叠加效应，也可能得益于用于数据记录的传感器数量增加（>550个传感器）。特别是，EMSI以无创方式定位了激惹区和SOZ，其结果与颅内脑电图（iEEG）金标准下的ESI结果一致，证实了临床观察结果。

本研究提出的方法包括以下关键步骤：（i）对药物难治性癫痫患儿，采集高质量的同步脑磁图（MEG）和高密度脑电图（HD-EEG）数据（即高信噪比），传感器空间采样密度高（>550个传感器），覆盖全脑，记录发作间期和发作期活动，以及视觉、运动、听觉和体感诱发的场与电位（步骤3.1–3.2）；（ii）对使用不同采集系统记录的MEG与HD-EEG信号进行时间同步和空间共配准（步骤3.12）；（iii）仔细预处理并分别选取包含发作间期活动（步骤4.1.1–4.1.7）、发作起始活动（步骤4.2.1–4.2.7）以及事件相关反应（步骤4.3.1–4.3.6）的数据片段；（iv）采用可靠的源定位方法（如结合聚类的等效偶极子模型ECDs和分布式源成像dSPM），精确进行激惹区、癫痫发作起始区（SOZ）以及感兴趣功能脑区的源定位（分别对应步骤4.1.8–4.1.9、4.2.8–4.2.9和4.3.7–4.3.9）。

进行同步MEG和高密度EEG记录时最关键的步骤，是对两个采集系统所记录的数据进行空间（坐标系之间的对齐）和时间（线性时钟漂移的校正）同步。这种同步对于准确识别在MEG和高密度EEG信号中同时发生的痫样放电、发作期事件以及视觉/运动/听觉/触觉事件至关重要。若这些事件的时间点选择出现误差，可能会影响脑电源定位结果，并错误识别出并未真正参与这些事件产生的脑区。

MEG 系统通常提供兼容的 32、64 和 128 导 EEG 系统，集成于产品中，用于同步进行 MEG 和 EEG 测量。在这种情况下，无需通过发送公共触发信号来实现数据的时间同步。同样，目前大多数 EEG 系统都可与所有 MEG 系统兼容。尽管硬件技术不断进步，但仅有少数癫痫中心在术前评估中常规开展同步 MEG 和高密度 EEG（HD-EEG）记录。本文利用了这种集成能力，结合了 306 导 MEG 和 256 导 EEG 系统，使用覆盖受试者头部的 >550 个传感器同步记录脑部活动。迄今为止，可用于 MEG、HD-EEG 和颅内 EEG（iEEG）数据高级分析的软件（例如 Brainstorm、CURRY、EEGLab、FieldTrip、MNE 或 NUTMEG）仍然较少。因此，未来研究有必要结合新的神经影像分析软件，进一步验证所提出的方法学。最后，将磁源成像（MSI）和电源成像（ESI）整合为统一解决方案（EMSI），增加了数据分析的计算复杂性。

所述方法存在一些局限性，需在今后的研究中加以解决。我们手动选取了两名代表性患者在MEG和高密度EEG数据中均出现的痫样放电事件（IEDs），而忽略了仅在其中一种信号（MEG或EEG）中出现的间歇期痫样放电。手动选取放电事件可能耗时较长且具有主观性，而过去几十年开发的用于检测IEDs的自动化方法可简化这一过程。57,58,59然而，仍建议通过视觉检查对每个痫样放电（IED）进行仔细分析和精细检测。此外，我们采用癫痫发作起始区（SOZ）作为癫痫灶（EZ）的近似指标。但SOZ并不总能预测手术预后效果。60,61,62,63因此，未来的研究可将手术结果作为确定癫痫灶（EZ）更精确边界的标准依据13,14,15,16,17,19,20尽管可以通过同步使用MEG和EEG成功捕获癫痫发作，并利用适当的源定位技术进行定位44,64在临床实践中，记录此类发作事件相对罕见，尤其是正在使用抗癫痫药物（ASMs）的门诊患者。这主要是由于脑磁图（MEG）记录时间有限，以及发作期间常出现大幅度的身体运动（例如患者头部滑出杜瓦瓶），这些运动可能产生生物伪迹，严重影响脑电源定位结果的准确性。在最近的一篇综述中，Stefan 等人. 报道了在不同研究中，7%至24%的患者在进行MEG记录期间出现癫痫发作，记录时间从平均30分钟至最长5.7小时不等65. 在CCMC，过去约两年内接受同步MEG与HD-EEG记录的89例患者中，有18例（20.2%）捕捉到发作事件。然而，仅有8例（占18例中的44.4%）患者成功完成分析。当发作间期MEG记录显示正常或结果不明确时，发作期MEG或HD-EEG可用于高精度定位癫痫灶（EZ）。然而，此类记录的技术与操作要求需予以充分考虑。此外，通过EMSI实现的功能区皮层定位代表性数据尚未与任何功能脑区定位的金标准（如无创fMRI或术中皮层电刺激）进行比较。因此，未来研究可整合EMSI与fMRI，开发一种多模态无创成像工具，以提高对药物难治性癫痫（DRE）患儿功能区脑区的定位准确性。该方法还可拓展至其他功能脑区的定位，例如语言功能区。在DRE患者的术前评估中，语言功能的定位至关重要，有助于判断患者是否适合手术、规划手术切除范围，并预防术后永久性功能缺损。66多项非侵入性研究表明，使用MEG进行语言定位 mapping 可提供与侵入性Wada试验相似的一致性结果，后者通常被视为识别优势语言半球的金标准。67,68,69,70最近一项研究提出了一种多模态方法，通过结合不同技术（即皮层电刺激定位、高伽马波段皮层脑电图、功能性磁共振成像和经颅磁刺激），可为术前语言功能区定位提供相互验证和互补的信息71尽管具有这些优势，由于年龄相关的认知、智力和语言障碍，儿童患者中的语言区定位仍然具有挑战性。因此，在不久的将来应开发更多适合特定年龄段且对儿童友好的任务范式和实验设置。在本研究中，我们使用的软件尚未获得用于临床目的的认证，尽管这些工具已被证明具有较高的价值和有效性，但在报告用于临床手术前评估的结果时，仍需考虑其潜在的责任问题。本文描述了仅使用基于海绵电极的高密度脑电图（HD-EEG）系统进行记录的操作流程。基于导电膏电极的EEG系统在临床和科研环境中也被广泛使用。虽然这类系统可提供信噪比更高的EEG记录，但其准备时间较长（约40–60分钟），因此在儿科应用中适用性较低。此外，一些实验室在进行脑磁图（MEG）记录时采用低密度导电膏EEG系统，这类系统在准备时间方面优于HD-EEG系统，但由于覆盖全头皮的电极数量较少，其空间分辨率显著降低。12,16,72,73.

目前，癫痫患者致痫脑区的定位主要仍依靠颅内脑电图（iEEG）监测。此外，对于功能重要脑区的精确定位方法尚不明确，而现有的脑磁图（MEG）实验室所采用的实验装置不适合儿童患者，同时高密度脑电图（HD-EEG）在该领域的应用也非常有限。准确识别这些脑区有助于术前评估，并优化针对病灶切除或iEEG电极植入的手术规划。迄今为止，已有若干研究探讨了脑电图源成像（ESI）或脑磁图源成像（MSI）在药物难治性癫痫（DRE）及局灶性癫痫患者术前评估中的作用，用于定位癫痫灶（EZ）12,13,14,15,16,17,18,19 以及体感皮层等重要功能区41。少数研究表明，相较于单独使用MSI或ESI，联合源成像（EMSI）在源定位精度和预后预测性能方面表现更优13,31,42。尽管有上述研究进展，MEG与EEG同步记录仍极少开展，MSI和ESI技术也仅在全球少数癫痫中心得以应用。据我们所知，本研究是首个为儿童癫痫患者提供同步采集与分析MEG和HD-EEG数据、并实施EMSI以无创识别激惹区、发作起始区（SOZ）以及包括初级视觉、运动、听觉和体感皮层在内的功能重要脑区的方法建议的研究。

本研究中，我们对两名药物难治性癫痫（DRE）患者（病例1和病例2）同步采集的无创数据中检测到的发作间期棘波和发作期事件进行了EMSI分析，得到的源定位误差分别约为9 mm和12 mm（相对于致痫区SOZ），结果与既往研究一致42。值得注意的是，该方法的定位精度可与颅内记录结果（即基于颅内脑电图iEEG数据的ESI）相媲美，所获得的偶极子聚集于临床观察所确定的致痫脑区（图3C 和 图4B）。此外，我们利用第三名典型DRE患者（病例3）的无创数据，对视觉、运动、听觉和体感诱发电活动进行了EMSI分析，结果在相应的功能脑区（即视觉皮层、运动皮层、听觉皮层和体感皮层）观察到显著的源激活模式（图5C、 图6C、图7C、 和 图8C）。

我们的结果源于脑磁图（MEG）和脑电图（EEG）两种模态所获取的互补信息的融合，这种融合可能提高定位的准确性。众所周知，脑电图能够反映所有颅内电流，而脑磁图主要对切向源敏感，对深部脑区源不敏感29,74。本研究结果表明，结合MEG和EEG可克服各自模态的局限性，提供更优的定位结果，并识别出单独使用脑电图源成像（ESI）或脑磁图源成像（MSI）时可能遗漏的致痫区和功能重要脑区。此外，对于术前评估期间未接受功能磁共振成像（fMRI）检查的患者，我们提出了一种利用脑磁图-脑电图联合源成像（EMSI）非侵入性定位功能重要脑区的替代方法。

使用无创技术（如同步 MEG 和 EEG）对致痫区及功能重要脑区进行定位，是术前评估药物难治性癫痫（DRE）患儿的重要步骤，旨在完全切除或断开癫痫灶（EZ）的同时保护功能皮层区域。本方案详细描述了同步 MEG 和 EEG 数据的采集与分析方法，不仅适用于癫痫手术前评估，还可应用于认知神经科学领域，用于研究正常发育儿童及健康成人健康大脑的生理功能，以及与癫痫或其他神经系统疾病相关的脑形态和功能改变。未来针对致痫脑网络的研究还可进一步探讨：基于同步 MEG 和高密度 EEG（HD-EEG）数据，通过 EMSI 无创估算的网络枢纽（即高度连接的脑区）是否比单独使用 MSI 和/或 ESI 方法能更准确地定位 DRE 患儿的 EZ75,76,77。此外，通过 EMSI 估算的尖波和涟波（即高频振荡，>80 Hz）在时空上的传播模式，可实现对癫痫样活动传播的病理生理机制进行更深入的理解，并无创评估这些传播活动的起始发生器——这是 EZ 的精确生物标志物78,79。本方案有助于进一步研究 MEG 与 EEG 系统之间的互补性，例如通过分析 MEG 和 EEG 传感器阵列对不同方向源信号的敏感性，从而在同步进行 MEG 和 HD-EEG 记录时，增进对脑电生理特性的理解。

致谢

$$\rightleftharpoonup{xx}$$ $$\longleftharp{xx}$$, $$\longrightharp{xx}$$,

本研究由国家神经疾病与卒中研究所（R01NS104116；R01NS134944；主要研究者：Christos Papadelis）资助。

材料

本文使用的材料清单
姓名公司目录编号评论
AIRSTIM 设备SD InstrumentsN/ASDI AIRSTIM 系统是一种替代电击的非条件刺激手段
婴儿洗发水 Johnson'sN/A婴儿洗发水对眼睛的温和程度堪比纯净水，专为轻柔清洁婴儿’娇嫩的头发和头皮而设计。 
Control III 消毒清洁液Maril Products, Inc.http://www.controlthree.com/专为医院配制的消毒剂和杀菌溶液
Elekta NeuromagTRIUXNM24132A一种综合性生物电磁测量系统，配备306通道神经磁强计，用于功能性脑研究
FASTRAKPolhemus 技术NS-7806FASTRAK 采用交流电磁技术，可提供精确的位置和方向数据，几乎无延迟。通过单一磁源，FASTRAK 可同时为最多四个传感器提供数据。磁源发射电磁场，处于有效范围内的传感器可被完整追踪其六自由度（6DOF）运动。系统设置简单直观，无需用户校准。
Genuine Grass 可重复使用杯状脑电电极Natus Medical, Inc.N/A每枚 Genuine Grass 脑电电极均经过严格机械和电气测试，确保长久使用寿命，实现无与伦比的记录清晰度和可靠性
Geodesic 传感器网Electrical Geodesics, Inc.S-MAN-200-GSNR-001包含32至256个电极，可置于人头部以采集高密度阵列脑电图数据
GeoScan 传感器数字化系统Electrical Geodesics, Inc.8100550-03手持式扫描仪及软件，用于三维电极位置注册
Natus Xltek NeuroWorksNatus Medical, Inc.https://natus.com/Natus NeuroWorks 平台简化了常规脑电图检测、动态脑电图、长期监测、重症监护病房监测及科研研究中的数据采集、监控与管理流程
Natus NeuroWorks 脑电图软件Natus Medical, Inc.https://natus.com/neuro/neuroworks-eeg-software/NeuroWorks 脑电图软件简化了脑电图检测数据的采集、监控、趋势分析与管理流程，使医护人员能够节省时间，专注于提供最佳诊疗服务
ROSA ONE BrainZimmer Biomethttps://www.zimmerbiomet.com/en/products-and-solutions/zb-edge/robotics/rosa-brain.htmlROSA ONE Brain 是一种机器人解决方案，可协助外科医生通过颅骨上的微小钻孔规划并执行复杂的神经外科手术 
Ten20 导电膏Weaver and companyN/ATen20 含有适当的黏附性与导电性平衡成分，可使电极稳固贴合，同时允许电信号有效传导

参考文献

$$\rightleftharpoonup{xx}$$ $$\longleftharp{xx}$$, $$\longrightharp{xx}$$,
  1. Oldham, M. S., Horn, P. S., Tsevat, J., Standridge, S. Costs and clinical outcomes of epilepsy surgery in children with drug-resistant epilepsy. Pediatr Neurol. 53 (3), 216-220 (2015).
  2. Willie, J. T., et al. Real-time magnetic resonance-guided stereotactic laser amygdalohippocampotomy for mesial temporal lobe epilepsy. Neurosurgery. 74 (6), 569-584 (2014).
  3. Rosenow, F., Lüders, H. Presurgical evaluation of epilepsy. Brain. 124, Pt 9 1683-1700 (2001).
  4. Önal, Ç, et al. Complications of invasive subdural grid monitoring in children with epilepsy). J Neurosurg. 98 (5), 1017-1026 (2003).
  5. Hader, W. J., et al. Complications of epilepsy surgery-a systematic review of focal surgical resections and invasive EEG monitoring. Epilepsia. 54 (5), 840-847 (2013).
  6. Meng, Y., et al. Risk factors for surgical site infection after intracranial electroencephalography monitoring for epilepsy in the pediatric population. J Neurosurg Pediatr. 22 (1), 31-36 (2018).
  7. Jobst, B. C., et al. Intracranial EEG in the 21st Century. Epilepsy Curr. 20 (4), 180-188 (2020).
  8. Schwartz, E. S., et al. Magnetoencephalography for pediatric epilepsy: how we do it. AJNR Am J Neuroradiol. 29 (5), 832-837 (2008).
  9. Michel, C. M., et al. Electric source imaging of human brain functions. Brain Res Rev. 36 (2-3), 108-118 (2001).
  10. Michel, C. M., He, B. EEG source localization. Handb Clin Neurol. 160, 85-101 (2019).
  11. Michel, C. M., Brunet, D. EEG source imaging: A practical review of the analysis steps. Front Neurol. 10, 325(2019).
  12. Plummer, C., et al. Interictal and ictal source localization for epilepsy surgery using high-density EEG with MEG: a prospective long-term study. Brain. 142 (4), 932-951 (2019).
  13. Duez, L., et al. Electromagnetic source imaging in presurgical workup of patients with epilepsy: A prospective study. Neurology. 92 (6), e576-e586 (2019).
  14. Mouthaan, B. E., et al. Diagnostic accuracy of interictal source imaging in presurgical epilepsy evaluation: A systematic review from the E-PILEPSY consortium. Clin Neurophysiol. 130 (5), 845-855 (2019).
  15. Pellegrino, G., et al. Clinical yield of magnetoencephalography distributed source imaging in epilepsy: A comparison with equivalent current dipole method. Hum Brain Mapp. 39 (1), 218-231 (2018).
  16. Tamilia, E., et al. Assessing the localization accuracy and clinical utility of electric and magnetic source imaging in children with epilepsy. Clin Neurophysiol. 130 (4), 491-504 (2019).
  17. Coito, A., et al. Interictal epileptogenic zone localization in patients with focal epilepsy using electric source imaging and directed functional connectivity from low-density EEG. Epilepsia Open. 4 (2), 281-292 (2019).
  18. Singh, J., Ebersole, J. S., Brinkmann, B. H. From theory to practical fundamentals of electroencephalographic source imaging in localizing the epileptogenic zone. Epilepsia. 63 (10), 2476-2490 (2022).
  19. Brodbeck, V., et al. Electrical source imaging for presurgical focus localization in epilepsy patients with normal MRI. Epilepsia. 51 (4), 583-591 (2010).
  20. Santalucia, R., et al. Clinical added value of interictal automated electrical source imaging in the presurgical evaluation of MRI-negative epilepsy: A real-life experience in 29 consecutive patients. Epilepsy Behav. 143, 109229(2023).
  21. Schneider, F., et al. Magnetic source imaging in non-lesional neocortical epilepsy: additional value and comparison with ICEEG. Epilepsy Behav. 24 (2), 234-240 (2012).
  22. Hämäläinen, M., Hari, R., Ilmoniemi, R. J., Knuutila, J., Lounasmaa, O. V. Magnetoencephalography-theory, instrumentation, and applications to noninvasive studies of the working human brain. Rev Mod Phys. 65, 413-497 (1993).
  23. Baillet, S., Mosher, J. C., Leahy, R. M. Electromagnetic brain mapping. IEEE Signal Process Mag. 18 (6), 14-30 (2001).
  24. Fuchs, M., Kastner, J., Tech, R., Wagner, M., Gasca, F. MEG and EEG dipole clusters from extended cortical sources. Biomed Eng Lett. 7 (3), 185-191 (2017).
  25. Singh, S. P. Magnetoencephalography: Basic principles. Ann Indian Acad Neurol. 17, Suppl 1 S107-S112 (2014).
  26. Ahlfors, S. P., Han, J., Belliveau, J. W., Hämäläinen, M. S. Sensitivity of MEG and EEG to source orientation. Brain Topogr. 23, 227-232 (2010).
  27. Kim, H., Chung, C. K., Hwang, H. Magnetoencephalography in pediatric epilepsy. Korean J Pediatr. 56 (10), 431-438 (2013).
  28. Gorjan, D., Gramann, K., De Pauw, K., Marusic, U. Removal of movement-induced EEG artifacts: Current state of the art and guidelines. J Neural Eng. , (2022).
  29. Barkley, G. L., Baumgartner, C. MEG and EEG in epilepsy. J Clin Neurophysiol. 20 (3), 163-178 (2003).
  30. Ebersole, J. S., Ebersole, S. M. Combining MEG and EEG source modeling in epilepsy evaluations. J Clin Neurophysiol. 27 (6), 360-371 (2010).
  31. Yoshinaga, H., et al. Benefit of simultaneous recording of EEG and MEG in dipole localization. Epilepsia. 43 (8), 924-928 (2002).
  32. Baumgartner, C. Controversies in clinical neurophysiology. MEG is superior to EEG in the localization of interictal epileptiform activity: Con. Clin Neurophysiol. 115 (5), 1010-1020 (2004).
  33. Barkley, G. L. Controversies in neurophysiology. MEG is superior to EEG in localization of interictal epileptiform activity: Pro. Clin Neurophysiol. 115 (5), 1001-1009 (2004).
  34. Braeutigam, S. Magnetoencephalography: Fundamentals and established and emerging clinical applications in radiology. ISRN Radiol. 2013, 529463(2013).
  35. Papadelis, C., et al. Current and emerging potential for magnetoencephalography in pediatric epilepsy. J Pediatr Epilepsy. 2 (1), 73-85 (2013).
  36. Fiedler, P., Fonseca, C., Supriyanto, E., Zanow, F., Haueisen, J. A high-density 256-channel cap for dry electroencephalography. Hum Brain Mapp. 43 (4), 1295(2022).
  37. Sharon, D., Hämäläinen, M. S., Tootell, R. B. H., Halgren, E., Belliveau, J. W. The advantage of combining MEG and EEG: comparison to fMRI in focally-stimulated visual cortex. Neuroimage. 36 (4), 1225(2007).
  38. Pataraia, E., Lindinger, G., Deecke, L., Mayer, D., Baumgartner, C. Combined MEG/EEG analysis of the interictal spike complex in mesial temporal lobe epilepsy. Neuroimage. 24 (3), 607-614 (2005).
  39. Ahmed Mahmutoglu, M., Rupp, A., Baumgärtner, U. Simultaneous EEG/MEG yields complementary information of nociceptive evoked responses. Clin Neurophysiol. 143, 21-35 (2022).
  40. Aydin, Ü, et al. Combined EEG/MEG can outperform single modality EEG or MEG source reconstruction in presurgical epilepsy diagnosis. PLoS One. 10 (3), e0118753(2015).
  41. Bast, T., et al. Combined EEG and MEG analysis of early somatosensory evoked activity in children and adolescents with focal epilepsies. Clin Neurophysiol. 118 (8), 1721-1735 (2007).
  42. Chikara, R. K., et al. Electromagnetic source imaging predicts surgical outcome in children with focal cortical dysplasia. Clin Neurophysiol. 153, 88-101 (2023).
  43. Hari, R., et al. IFCN-endorsed practical guidelines for clinical magnetoencephalography (MEG). Clin Neurophysiol. 129 (8), 1720-1747 (2018).
  44. Bagić, A. I., Knowlton, R. C., Rose, D. F., Ebersole, J. S. American Clinical Magnetoencephalography Society Clinical Practice Guideline 1: Recording and analysis of spontaneous cerebral activity. J Clin Neurophysiol. 28 (4), 348-354 (2011).
  45. Papadelis, C., Chen, Y. H. Pediatric magnetoencephalography in clinical practice and research. Neuroimaging Clin N Am. 30 (2), 239-248 (2020).
  46. De Macedo Rodrigues, K., et al. A FreeSurfer-compliant consistent manual segmentation of infant brains spanning the 0-2 year age range. Front Hum Neurosci. 9, 21(2015).
  47. Tadel, F., Baillet, S., Mosher, J. C., Pantazis, D., Leahy, R. M. Brainstorm: A user-friendly application for MEG/EEG analysis. Comput Intell Neurosci. 2011, 879716(2011).
  48. Kane, N., et al. A revised glossary of terms most commonly used by clinical electroencephalographers and updated proposal for the report format of the EEG findings. Revision 2017. Clin Neurophysiol Pract. 2, 170-185 (2017).
  49. Lantz, G., et al. Propagation of interictal epileptiform activity can lead to erroneous source localizations: a 128-channel EEG mapping study. J Clin Neurophysiol. 20 (5), 311-319 (2003).
  50. Vorwerk, J., et al. A guideline for head volume conductor modeling in EEG and MEG. Neuroimage. 100, 590-607 (2014).
  51. Schrader, S., et al. DUNEuro-A software toolbox for forward modeling in bioelectromagnetism. PLoS One. 16 (6), e0252431(2021).
  52. Ntolkeras, G., et al. Presurgical accuracy of dipole clustering in MRI-negative pediatric patients with epilepsy: Validation against intracranial EEG and resection. Clin Neurophysiol. 141, 126-138 (2022).
  53. David, O., Kilner, J. M., Friston, K. J. Mechanisms of evoked and induced responses in MEG/EEG. Neuroimage. 31, 1580-1591 (2006).
  54. Pantev, C. Evoked and induced gamma-band activity of the human cortex. Brain Topogr. 7 (4), 321-330 (1995).
  55. Fox, N. A., et al. Assessing human mirror activity with EEG mu rhythm: A meta-analysis. Psychol Bull. 142 (3), 291-313 (2016).
  56. Genzer, S., Ong, D. C., Zaki, J., Perry, A. Mu rhythm suppression over sensorimotor regions is associated with greater empathic accuracy. Soc Cogn Affect Neurosci. 17 (9), 788-801 (2022).
  57. Janmohamed, M., et al. Moving the field forward: detection of epileptiform abnormalities on scalp electroencephalography using deep learning-clinical application perspectives. Brain Commun. 4 (5), 218(2022).
  58. Bagheri, E., Jin, J., Dauwels, J., Cash, S., Westover, M. B. A fast machine learning approach to facilitate the detection of interictal epileptiform discharges in the scalp electroencephalogram. J Neurosci Methods. 326, 108362(2019).
  59. Thomas, J., et al. Automated detection of interictal epileptiform discharges from scalp electroencephalograms by convolutional neural networks. Int J Neural Syst. 30 (11), 2050030(2020).
  60. Zijlmans, M., Zweiphenning, W., van Klink, N. Changing concepts in presurgical assessment for epilepsy surgery. Nat Rev Neurol. 15 (10), 594-606 (2019).
  61. Akiyama, T., et al. Focal resection of fast ripples on extraoperative intracranial EEG improves seizure outcome in pediatric epilepsy. Epilepsia. 52 (10), 1802-1811 (2011).
  62. Duncan, J. S., Winston, G. P., Koepp, M. J., Ourselin, S. Brain imaging in the assessment for epilepsy surgery. Lancet Neurol. 15 (4), 420(2016).
  63. Jacobs, J., et al. High-frequency electroencephalographic oscillations correlate with outcome of epilepsy surgery. Ann Neurol. 67 (2), 209-220 (2010).
  64. Ricci, L., et al. Virtual implantation using conventional scalp EEG delineates seizure onset and predicts surgical outcome in children with epilepsy. Clin Neurophysiol. 139, 49-57 (2022).
  65. Stefan, H., Rampp, S. Interictal and Ictal MEG in presurgical evaluation for epilepsy surgery. Acta Epileptologica. 2, 11(2020).
  66. Jahangiri, F. R., Chima, G. S., Pearson, M., Jackson, J., Siddiqui, A. A. Mapping of the language cortex. Cureus. 13 (5), e14960(2021).
  67. Merrifield, W. S., Simos, P. G., Papanicolaou, A. C., Philpott, L. M., Sutherling, W. W. Hemispheric language dominance in magnetoencephalography: sensitivity, specificity, and data reduction techniques. Epilepsy Behav. 10 (1), 120-128 (2007).
  68. Wheless, J. W., et al. Magnetoencephalography (MEG) and magnetic source imaging (MSI). Neurologist. 10 (3), 138-153 (2004).
  69. Breier, J. I., Simos, P. G., Zouridakis, G., Papanicolaou, A. C. Lateralization of activity associated with language function using magnetoencephalography: a reliability study. J Clin Neurophysiol. 17 (5), 503-510 (2000).
  70. Pataraia, E., Baumgartner, C., Lindinger, G., Deecke, L. Magnetoencephalography in presurgical epilepsy evaluation. Neurosurg Rev. 25 (3), 141-159 (2002).
  71. Babajani-Feremi, A., et al. Language mapping using high gamma electrocorticography, fMRI, and TMS versus electrocortical stimulation. Clin Neurophysiol. 127 (3), 1822-1836 (2016).
  72. Brodbeck, V., et al. Electroencephalographic source imaging: a prospective study of 152 operated epileptic patients. Brain. 134, Pt 10 2887-2897 (2011).
  73. Sohrabpour, A., et al. Effect of EEG electrode number on epileptic source localization in pediatric patients. Clin Neurophysiol. 126 (3), 472-480 (2015).
  74. Laohathai, C., et al. Practical fundamentals of clinical MEG interpretation in epilepsy. Front Neurol. 12, 722986(2021).
  75. Corona, L., et al. Mapping functional connectivity of epileptogenic networks through virtual implantation. Proceedings of the Annu Int Conf IEEE Eng Med Biol Soc. 2021, 408-411 (2021).
  76. Corona, L., et al. Non-invasive mapping of epileptogenic networks predicts surgical outcome. Brain. 146 (5), 1916-1931 (2023).
  77. Rijal, S., et al. Functional connectivity discriminates epileptogenic states and predicts surgical outcome in children with drug resistant epilepsy. Sci Rep. 13 (1), 9622(2023).
  78. Tamilia, E., et al. Noninvasive mapping of ripple onset predicts outcome in epilepsy surgery. Ann Neurol. 89 (5), 911-925 (2021).
  79. Matarrese, M. A. G., et al. Spike propagation mapping reveals effective connectivity and predicts surgical outcome in epilepsy. Brain. 146 (9), 3898-3912 (2023).

重印与许可

申请许可以重复使用本 JoVE 文章的文本或图表

申请许可

标签

MEG

相关文章