MEG versus EEG localization test-猫眼课题宝

MEG 与 EEG 定位测试

DOI：

10.1002/ana.410300221

发表时间：

1991

期刊：

Annals of Neurology

影响因子：

11.2

通讯作者：

中科院分区：

医学1区

文献类型：

作者：

R. Hari;Matti Hämäläinen;R. Ilmoniemi;O. V. Lounasmaa

在最近的一份报告中，Cohen及其同事E1比较了植入患者大脑中的源的信号的电磁测量结果，并得出结论，结果表明脑磁图（MEG）在定位局灶性源方面并没有比脑电图（EEG）提供显着的优势。不幸的是，科恩及其同事的MEG测量不符合当代标准，因为他们是在不适当的仪器和过时的方法进行的，这严重限制了他们结果的可靠性。当在头皮上连续获得现场测量时，在拟合源参数时显然没有考虑传感器的方向。此故障可能会在定位源时引入严重错误。仪器可用于指定传感器的方向，并且与Cohen及其同事报告的8 mm误差相比，包含这种设备的程序可以在模型球体或颅骨内约4 cm深度处定位源时提供优于3 mm的一致精度[2]。事实上，Cohen及其同事报告了类似的3 mm精度，用于安装在导电球体内的强电流偶极子的测量，其中很容易径向对准传感器。他们研究的第二个主要限制是在整个田间模式中取样不足。在噪声存在的情况下，重要的是要在磁场极值的所有侧面获得测量值，即径向磁场在离开和进入磁头时最强的地方。在文章[l]的图2中可以清楚地看到患者1缺乏适当的数据，其中在后极值和右前区域附近获得的测量值非常少。最后，应该注意的是，使用rf SQUID来感测磁场，而不是通常提供至少3倍或更低噪声的较新的dc SQUID。虽然作者没有解释是什么特定因素限制了他们的数据质量，但很明显，较低的噪声传感器为宽带检测提供了主要优势，如果受试者噪声的罕见事件占主导地位，则提供了选择性平均的可能性。由于这些严重的局限性，这项研究不应被认为是在人类大脑内定位磁源的最先进的代表。事实上，从基本的考虑来看，没有理由笼统地说EEG和MEG提供了等同的准确性，或者说一个上级另一个而没有资格。该优势在很大程度上取决于感兴趣的神经元源的位置和方向。过去十年的大量研究表明，MEG和EEG是互补的技术[3]。在某些情况下（例如，皮质源）MEG可以提供定位的优势，而在其他（脑干）EEG可能更有效。有一个方面是明确的：随着技术的改进，更大的传感器阵列的测量，头部几何形状的更准确的表征，和-特别是重要的EEG-解剖结构的电导率的改进表征，这两种技术将提供显着改善的定位和定量表征神经元活动。

In a recent report, Cohen and colleagues El1 compared electric and magnetic measurements of signals from a source implanted in a patient’s brain and concluded that the results suggest that the magnetoencephalogram (MEG) offers no significant advantage over the electroencephalogram (EEG) in localizing a focal source. Unfortunately, the MEG measurements of Cohen and associates d o not meet contemporary standards, because they were carried out with inadequate instrumentation and outdated methodology that severely limit the reliability of their results. When field measurements were obtained sequentially across the scalp, the orientation of their sensor apparently was not taken into account when fitting the source parameters. This failure can introduce serious errors in locating the source. Instrumentation is available to specify the sensor’s orientation, and procedures that incorporate such a device can provide a consistent accuracy of better than 3 mm in locating a source at a depth of about 4 cm within a model sphere or skull [2], compared with the 8-mm error reported by Cohen and associates. Indeed, Cohen and associates report a similar 3-mm accuracy for measurements with a strong current dipole mounted within a conducting sphere, where it is easy to align the sensor radially. A second major limitation in their study is inadequate sampling across the field pattern. In the presence of noise, it is imporrant to obtain measurements on all sides of the field extrema-where the radial field is strongest on leaving and on entering the head. The lack of appropriate data can be seen clearly for Patient 1 in Figure 2 of the article [l], where very few measurements were obtained near the posterior extremum and right anterior region. Finally, it should be noted that an rf SQUID was used to sense the magnetic field instead of the newer dc SQUID that typically provides at least a factor of 3 or lower noise. Although the authors did not explain what specific factor limited the quality of their data, it is clear that a lower noise sensor provides major advantages for wide-band detection, offering the possibility of selective averaging if infrequent events of subject noise dominate. Because of these serious limitations, this study should not be considered representative of the state-of-the-art in locating magnetic sources within the human brain. Indeed, from fundamental considerations there is no justification for a blanket statement that the EEG and MEG provide equivalent accuracy, or that one is superior to the other without qualifications. The advantage very much depends on the location and orientation of the neuronal source of interest. Evidence from numerous studies during the past decade indicates that the MEG and EEG are complementary techniques [3). In some situations (e.g., cortical sources) MEG may provide advantages in localization, while in others (brainstem) EEG may be more effective. One aspect is clear: With improved technology, measurements with larger sensor arrays, more accurate characterization of head geometries, and-especially important for the EEG-improved characterization of the electrical conductivity of anatomical structures, both techniques will provide significantly improved localization and quantitative characterization of neuronal activity.

Magnetic localization of neuronal activity in the human brain.