ATR Computational Neuroscience Laboratories, Kyoto 619-0288, Japan.
Neuroimage. 2012 Feb 15;59(4):4006-21. doi: 10.1016/j.neuroimage.2011.09.087. Epub 2011 Oct 14.
Previous simulation and experimental studies have demonstrated that the application of Variational Bayesian Multimodal EncephaloGraphy (VBMEG) to magnetoencephalography (MEG) data can be used to estimate cortical currents with high spatio-temporal resolution, by incorporating functional magnetic resonance imaging (fMRI) activity as a hierarchical prior. However, the use of combined MEG and fMRI is restricted by the high costs involved, a lack of portability and high sensitivity to body-motion artifacts. One possible solution for overcoming these limitations is to use a combination of electroencephalography (EEG) and near-infrared spectroscopy (NIRS). This study therefore aimed to extend the possible applications of VBMEG to include EEG data with NIRS activity as a hierarchical prior. Using computer simulations and real experimental data, we evaluated the performance of VBMEG applied to EEG data under different conditions, including different numbers of EEG sensors and different prior information. The results suggest that VBMEG with NIRS prior performs well, even with as few as 19 EEG sensors. These findings indicate the potential value of clinically applying VBMEG using a combination of EEG and NIRS.
先前的模拟和实验研究表明，通过将功能磁共振成像 (fMRI) 活动作为分层先验，将变分贝叶斯多模态脑图 (VBMEG) 应用于脑磁图 (MEG) 数据，可以用于以高时空分辨率估计皮质电流。然而，联合使用 MEG 和 fMRI 受到成本高、缺乏便携性和对身体运动伪影高度敏感等因素的限制。克服这些限制的一种可能方法是结合脑电图 (EEG) 和近红外光谱 (NIRS) 使用。因此，本研究旨在扩展 VBMEG 的可能应用范围，包括将 NIRS 活动作为分层先验的 EEG 数据。通过计算机模拟和真实实验数据，我们评估了 VBMEG 在不同条件下应用于 EEG 数据的性能，包括不同数量的 EEG 传感器和不同的先验信息。结果表明，即使使用 19 个 EEG 传感器，具有 NIRS 先验的 VBMEG 也能表现良好。这些发现表明，使用 EEG 和 NIRS 相结合的方法在临床上应用 VBMEG 具有潜在价值。
