The paper describes a multidisciplinary work that uses a model-based systems engineering method for developing real-time magnetoencephalography (MEG) signal processing. We introduce a requirement-driven, model-based development methodology (RDD and MBD) to provide a high-level environment and efficiently handle the complexity of computation and control systems. The proposed development methodology focuses on the use of System Modeling Language to define high-level model-based design descriptions for later implementation in heterogeneous hardware/software systems. The proposed approach was applied to the implementation of a real-time artifact rejection unit in MEG signal processing and demonstrated high efficiency in designing complex high-performance embedded systems. In MEG signal processing, biological artifacts in particular have a signal strength that overtop the signal of interest by orders of magnitude and must be removed from the measurement to achieve high-quality source reconstructions with minimal error contributions. However, many existing brain–computer interface studies overlook real-time artifact removal because of the demanding computational process. In this work, an automated real-time artifact rejection method is introduced, which is based on the recently presented method “ocular and cardiac artifact rejection for real-time analysis in MEG” (OCARTA). The method has been implemented using the RDD and MBD approach and successfully verified on a Virtex-6 field-programmable gate array.

本文描述了一项多学科的工作，该工作使用基于模型的系统工程方法来开发实时脑磁图（MEG）信号处理。我们引入了一种基于需求的，基于模型的开发方法（RDD和MBD），以提供一个高级环境并有效地处理计算和控制系统的复杂性。拟议的开发方法论着重于使用系统建模语言来定义基于高级模型的设计描述，以供以后在异构硬件/软件系统中实施。所提出的方法被应用于MEG信号处理中的实时伪像抑制单元的实现，并在设计复杂的高性能嵌入式系统中证明了高效率。在MEG信号处理中 尤其是生物伪像的信号强度要比目标信号高出几个数量级，因此必须从测量中去除，以实现误差贡献最小的高质量源重构。但是，由于要求很高的计算过程，许多现有的脑机接口研究都忽略了实时伪像去除。在这项工作中，介绍了一种自动实时伪像剔除方法，该方法基于最近提出的方法“用于MEG实时分析的眼部和心脏伪像剔除”（OCARTA）。该方法已使用RDD和MBD方法实现，并已在Virtex-6现场可编程门阵列上成功验证。