Wearable Magnetoencephalography (MEG) to detect brain activity.

我的研究项目旨在开发一种可穿戴（头盔或帽子）脑磁图（MEG）系统。脑磁图系统是一种医疗设备，用于通过磁传感器检测大脑活动。神经生理过程产生了电场和磁场。更常见的脑电图（EEG）系统使用电探针来检测神经元脉冲，但由于通过头骨和头皮的电场衰减，它无法提供大脑活动的精确空间分辨率。MEG已经被证明可以使用磁传感器提供更明确的信息。该项目旨在使用微型磁传感器，使其可穿戴，更易于在临床环境中使用。现有的MEG系统体积庞大，有特殊要求，如金属绝缘室以防止背景辐射。脑磁图的磁传感器也必须是极冷的，大约在零下270摄氏度左右，这导致了一个巨大而昂贵的系统，不能在小型全科医生的诊所（GP）中使用。帽内的隧道磁阻（TMR）传感器将检测大脑神经元产生的磁场，并将其转换为信号。这个信号会有噪声（无用信号的高干扰），不能用于任何应用。噪声将来自相邻的神经元、背景辐射和电线。因此，本项目的下一步是去除机器/系统中的噪声，以获得更清晰、更平滑的信号。这个过程被称为信号处理，其中任何信号都可以实时滤波以提供所需的信号。然而，并不是所有的噪音都可以被去除，通过使用一种机器学习算法，将学习从噪音中分离神经活动，将能够区分纯粹的大脑信号。项目的下一步需要不断测试算法，因为算法的准确性和处理速度取决于进行的测试次数。一旦获得纯粹的信号，神经学家将能够通过计算机看到实际的大脑活动，就像目前的MEG机器一样。拟议的研究也将有其他潜在的应用。例如，科学家已经测试了一个喂食机器人，通过识别大脑中负责运动功能的区域，在猴子的大脑中嵌入了一个探针。这只猴子能够通过脑机接口系统控制机械臂给它喂香蕉。因此，如果一种可穿戴式脑磁图帽能够被开发出来，人类将能够以比目前使用基于脑磁图的设备更好的精度，用他们的大脑控制机器人甚至假肢。创造一个精确的、无噪声的、便携的记录神经元活动的系统是在临床或家庭环境中广泛接受脑机接口系统的最重要的先决条件。

My research project aims to develop a wearable (helmet or cap) Magnetoencephalography (MEG) system. A MEG system is a medical device used to detect the brain's activity with magnetic sensors. Neurophysiological process creates both electrical and magnetic field. Where the more common, electroencephalography (EEG) system, uses electrical probes to detect electrical neuron pulses, it fails to provide a precise spatial resolution of the brains activities, due to attenuation of electrical filed through scull and scalp. MEG has already been proved to give much more explicit information using magnetic sensors. The project aims to use miniaturised magnetic sensors by making it wearable and more accessible to use in clinical environments. The existing MEG systems are bulky with special requirements, such as metal insulated rooms to prevent background radiation. The magnetic sensors of a MEG must also be extremely cold around -270 degrees Celsius and result in a huge and expensive system, which cannot be used in a small general practitioner's clinic (GP).The tunnel magnetoresistance (TMR) sensor in the cap will detect the generated magnetic field from the brain neurons and convert it into a signal. This signal will be noisy (high interference of unwanted signal) and cannot be used for any application. The noise will come from adjacent neurons, background radiation and wirings. Therefore, the next step in this project is to remove noise from the machine/system to give a clearer and smoother signal. This process is called signal processing, where any signal can be filtered in real time to provide the desired signal.However, not all noise can be removed and by using a machine-learning algorithm that will learn to separate neural activity from noise, will be able to distinguish a pure brain signal. This next step of the project requires the algorithm to be continually tested as the accuracy and process speed of the algorithm is depended on the number of tests conducted. Once a pure signal is obtained, a neurologist will be able to see the actual brain activity by computers, just like the current MEG machines.The proposed research will also have other potential applications. Such as scientists have already tested a feeding robot, whereby identifying the region of the brain responsible for motor functions, a probe was embedded in a monkey's brain. The monkey was able to control a robotic arm to feed it bananas via the brain-computer interface system. Therefore, if a wearable MEG cap can be developed, humans will be able to control robots or even prosthetics with their minds with much better precision than it is currently available using MEG-based devices. Creating a precise, noise free and portable systems for recording neuronal activity is the most important prerequisite towards wide spread acceptance of brain computer interface systems in clinical or home environments.