神经适应性脑电描记术:婴儿的原理验证研究
Neuroadaptive electroencephalography: a proof-of-principle study in infants
摘要:
功能性神经影像学的核心目标是研究大脑如何处理环境。主流范式涉及同时测量一个广泛的大脑反应的一小部分环境功能预选参考以前的研究或理论框架。作为补充,我们通过允许研究人员记录广泛的环境特征对预先选择的大脑反应的调制来反转这种方法。当理论框架或以前的经验数据贫乏时,我们的方法是最佳的。通过使用预先指定的闭环设计,该方法解决了大脑研究中再现性和普遍性的基本挑战。当研究发育中的大脑时,这些情况特别严重,我们基于成人大脑功能的理论可能从根本上歪曲了婴儿认知的地形,并且在数据采集方面存在重大的实际挑战。我们的方法采用机器学习来映射神经特征在实验刺激空间中的调制。我们的方法实时收集,处理和分析EEG大脑数据;并使用神经自适应贝叶斯优化算法来调整根据给定参与者的先前样本呈现的刺激。未采样的刺激可以通过沿着数据集拟合高斯过程回归来插值。我们表明,我们的方法可以自动识别的婴儿的母亲的脸,通过在线记录他们的NC大脑反应的脸连续。我们可以检索每个参与者的个性化响应的模型统计,为早期识别非典型发展打开了大门。这种方法在婴儿期研究及以后的研究中具有很大的潜力,可以提高绘制大脑功能的个体认知地形图的能力和普遍性。
英文摘要:
A core goal of functional neuroimaging is to study how the environment is processed in the brain. The mainstream paradigm involves concurrently measuring a broad spectrum of brain responses to a small set of environmental features preselected with reference to previous studies or a theoretical framework. As a complement, we invert this approach by allowing the investigator to record the modulation of a preselected brain response by a broad spectrum of environmental features. Our approach is optimal when theoretical frameworks or previous empirical data are impoverished. By using a prespecified closed-loop design, the approach addresses fundamental challenges of reproducibility and generalisability in brain research. These conditions are particularly acute when studying the developing brain, where our theories based on adult brain function may fundamentally misrepresent the topography of infant cognition and where there are substantial practical challenges to data acquisition. Our methodology employs machine learning to map modulation of a neural feature across a space of experimental stimuli. Our method collects, processes and analyses EEG brain data in real-time; and uses a neuro-adaptive Bayesian optimisation algorithm to adjust the stimulus presented depending on the prior samples of a given participant. Unsampled stimuli can be interpolated by fitting a Gaussian process regression along the dataset. We show that our method can automatically identify the face of the infant's mother through online recording of their Nc brain response to a face continuum. We can retrieve model statistics of individualised responses for each participant, opening the door for early identification of atypical development. This approach has substantial potential in infancy research and beyond for improving power and generalisability of mapping the individual cognitive topography of brain function.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。