Wang Fulong, Cao Fuzhi, Ma Yujie, Zhao Ruochen, Wang Ruonan, An Nan, Xiang Min, Wang Dawei, Ning Xiaolin
Key Laboratory of Ultra-Weak Magnetic Field Measurement Technology, Ministry of Education, School of Instrumentation and Optoelectronic Engineering, Beihang University, 100191, Beijing, China; Hangzhou Institute of Extremely-Weak Magnetic Field Major National Science and Technology Infrastructure, Hangzhou, 310051, China.
Key Laboratory of Ultra-Weak Magnetic Field Measurement Technology, Ministry of Education, School of Instrumentation and Optoelectronic Engineering, Beihang University, 100191, Beijing, China; Hangzhou Institute of Extremely-Weak Magnetic Field Major National Science and Technology Infrastructure, Hangzhou, 310051, China; School of Engineering Medicine, Beihang University, Beijing, 100191, China.
Neuroimage. 2025 Feb 1;306:120991. doi: 10.1016/j.neuroimage.2024.120991. Epub 2025 Jan 3.
Optically pumped magnetometer-based magnetoencephalography (OPM-MEG) is an novel non-invasive functional imaging technique that features more flexible sensor configurations and wearability; however, this also increases the requirement for environmental noise suppression. Subspace projection algorithms are widely used in MEG to suppress noise. However, in OPM-MEG systems with a limited number of channels, subspace projection methods that rely on spatial oversampling exhibit reduced performance. The homogeneous field correction (HFC) method resolves this problem by constructing a low-rank spatial model; however, it cannot address complex non-homogeneous noise. The spatiotemporal extended homogeneous field correction (teHFC) method uses multiple orthogonal projections to suppress disturbances. However, the signal and noise subspace are not completely orthogonal, limiting enhancement in the capabilities of the teHFC. Therefore, we propose an extended homogeneous field correction method based on oblique projection (opHFC), which overcomes the issue of non-orthogonality between the signal and noise subspace, enhancing the ability to suppress complex interferences. The opHFC constructs an oblique projection operator that divides the signals into internal and external components, eliminating complex interferences through temporal extension. We compared the opHFC with four benchmark methods by simulations and auditory and somatosensory evoked OPM-MEG experiments. The results demonstrate that opHFC provides superior noise suppression with minimal distortion, enhancing the signal quality at the sensor and source levels. Our method offers a novel approach to reducing interference in OPM-MEG systems, expanding their application scenarios, and providing high-quality signals for scientific research and clinical applications based on OPM-MEG.
基于光泵磁力仪的脑磁图（OPM-MEG）是一种新型的非侵入性功能成像技术，具有更灵活的传感器配置和可穿戴性；然而，这也增加了对环境噪声抑制的要求。子空间投影算法在脑磁图中被广泛用于抑制噪声。然而，在通道数量有限的OPM-MEG系统中，依赖空间过采样的子空间投影方法性能会降低。均匀场校正（HFC）方法通过构建低秩空间模型解决了这个问题；然而，它无法处理复杂的非均匀噪声。时空扩展均匀场校正（teHFC）方法使用多个正交投影来抑制干扰。然而，信号子空间和噪声子空间并非完全正交，限制了teHFC能力的提升。因此，我们提出了一种基于斜投影的扩展均匀场校正方法（opHFC），它克服了信号子空间和噪声子空间之间的非正交性问题，增强了抑制复杂干扰的能力。opHFC构建了一个斜投影算子，将信号分为内部和外部分量，通过时间扩展消除复杂干扰。我们通过模拟以及听觉和体感诱发的OPM-MEG实验，将opHFC与四种基准方法进行了比较。结果表明，opHFC在最小失真的情况下提供了卓越的噪声抑制能力，提高了传感器和源水平的信号质量。我们的方法为减少OPM-MEG系统中的干扰提供了一种新方法，扩展了其应用场景，并为基于OPM-MEG的科学研究和临床应用提供了高质量的信号。
