低频高分辨率大脑磁感应成像技术研究

The low frequency (below 10 kHz) electrical properties of brain tissues are important for the brain research such as of EEG/MEG source imaging and transcranial magnetic stimulation. In principle, low frequency magnetic induction tomography (MIT) can use an induction coil array to noninvasively map the electrical properties of brain tissues without contact. Due to the limitation of strong direct coupling of the source coils and the long wavelength, however, the low frequency scattering signals (secondary fields) produced by biological tissues are very low compared to the direct coupling, thus leading to a poor signal-to-noise ratio and low spatial resolution in MIT mapping of the electrical properties. Hence, how to improve the relative strength of the useful signals, the spatial resolution and accuracy is one of the hot topics in MIT research. With the goal of achieving low frequency high spatial resolution MIT technology for brain imaging, this project will carry out the following research: (1) Develop a new bucking coil technique to decouple the direct field and thus improve the signal level of the secondary fields. (2) Develop the forward and inverse solvers for low-frequency MIT. (3) Use the high spatial resolution magnetic resonance electrical properties tomography (MREPT) to constrain MIT and thus achieve low frequency high spatial resolution MIT, and perform phantom and animal tests.

10kHz以下的低频大脑电特性分布对脑电成像和经颅磁刺激研究极为重要但难以高精度实测。低频磁感应成像技术在原理上可以利用感应线圈阵列无接触、无侵入地对大脑进行电特性成像。但由于近场线圈的强耦合及低频电磁波波长特别长，低频下人脑散射的二次场远比直耦信号小，导致磁感应成像无法在低频频段获得准确的高分辨大脑电特性分布。如何提高磁感应信号质量、成像分辨率和精度是磁感应成像研究的热点问题。因此，本课题将围绕实现低频高分辨率大脑磁感应成像技术的目标开展以下研究：(1)研制新颖的补偿线圈去耦方法，提高低频磁感应成像系统对二次场的检测能力。(2)开发低频磁感应成像正演和反演算法。(3)利用磁共振的高分辨率电特性成像技术作为磁感应成像的先验信息，实现低频高分辨磁感应成像，并进行体模和动物实验。

作为一种无侵入式的成像方法，磁感应成像借助于磁场可以更好穿透大脑，定量地反演出大脑电特性。但受人体低电导率影响，低频下人体散射的二次场较弱，导致磁感应成像无法在低频频段获得准确的大脑电特性分布，特别是小于10 kHz频带。此外，低分辨率问题也是磁感应成像研究的热点问题之一。而利用高分辨率的磁共振电特性成像技术作为磁感应成像的先验信息可以解决这个问题。因此本课题围绕低频高分辨率大脑磁感应成像技术开展研究。首先，提出混合有限元法解决低频磁感应成像的低频崩溃问题，为反演计算提供一个精确的正演仿真软件；其次，构建半连接反向传播神经网络和U型网络为基础的反演算法，用于精确、高效地求解磁感应成像的反演问题；然后，以最大抵消原理为基础，设计磁感应成像实验系统，成功消除一次磁场的影响，获得准确的二次磁场；最后，改进神经网络反演算法，提高磁共振电特性成像的精度和效率，为磁感应成像提供精确的先验信息。