面向疼痛个体差异的脑电机制与解码方法研究

Accurate pain assessment is critical to pain diagnosis and management. Because pain is a subjective first-person experience, self-report is the gold standard for pain assessment in clinical practice. However, self-report of pain is not always available and could be biased, so physiological assessment of pain that bypasses self-report would be of greatly clinical significance. Recently, decoding pain perception from electroencephalography (EEG) has attracted a growing interest for it has the potential to provide objective and accurate assessment of pain. However, substantial individual differences (including inter- and intra-individual differences) in subjective pain perception and pain-related EEG activities significantly lower the accuracy and reliability of EEG-based pain decoding technique and limit its clinical practice. This project is aimed to characterize the inter- and intra-individual differences in EEG-based pain decoding models, to investigate the neural mechanisms of these individual differences, and to develop accurate and effective pain decoding models that can overcome the adverse effects of inter- and intra-individual differences. First, we will develop new EEG analyses methods to estimate and extract pain-related spatio-temporal-spectral EEG features on a trial-by-trial basis and will quantify the inter- and intra-individual differences in pain-related EEG features. Second, we will use EEG and multimodal magnetic resonance imaging (MRI) techniques to jointly investigate the neural mechanisms of individual differences in EEG-based pain decoding models and will identify EEG signatures of these individual differences. Third, we will develop and validate new EEG-based pain decoding models based on advanced machine learning methods and will use the EEG signatures of individual differences to guide the model development so that the new EEG-based pain decoding models can achieve high accuracy and reliability in cross-individual and cross-trial pain prediction. This project is expected to improve our understanding of the mechanisms of individual differences in pain experience and will pave the way for a precise, objective, and individualized pain assessment tool.

准确的疼痛测量是疼痛诊疗的关键前提。临床疼痛测量主要依赖患者主观评分，其可靠性和可行性有局限。从脑电中解码疼痛是当前被重点关注的一项有潜力客观精确测量疼痛的技术，但疼痛及其诱发脑电呈现的巨大个体差异（包括个体内差异和个体间差异）显著降低了该技术的准确性和适用性。本项目的目标是揭示疼痛脑电解码中个体差异的特性，探索个体差异的神经机制，并发展克服个体差异的解码方法。拟首先发展可精确估计疼痛诱发脑电差异性信息的特征提取算法，全面准确描述脑电特征的个体差异特性。其次，将融合脑电和磁共振影像探究疼痛解码个体内差异和个体间差异的机制，识别这两类个体差异的高时空分辨率脑电标记。最后，拟基于先进机器学习算法发展解码模型，并利用个体差异的脑电标记优化模型设计，使其可以克服个体差异影响，更准确可靠地解码疼痛。本项目将增强对大脑感知疼痛模式及其个体差异的理解，为推进疼痛脑电解码的临床转化提供理论基础和技术支持。

准确的疼痛测量是疼痛诊疗的关键前提。临床疼痛测量主要依赖患者主观评分，其可靠性和可行性有局限。从脑电中解码疼痛是当前被重点关注的一项有潜力客观精确测量疼痛的技术，但疼痛及其诱发脑电呈现的巨大个体差异显著降低了该技术的准确性和适用性。本项目的目标是揭示疼痛脑电解码中个体差异的特性，探索个体差异的神经机制，并发展克服个体差异的解码方法。我们首先联合国内多家单位共同建立了一个疼痛多模态神经信息数据库。该疼痛数据库采集超过400名被试在接受疼痛刺激任务前后和期间的数据，包括疼痛评分等行为学数据、脑电、结构与功能磁共振影像、基因数据等。该疼痛数据库的建立为本项目后续开展机制与算法研究打下基础。基于数据库，我们根据研究方案展开了三方面的工作：识别疼痛个体差异特征，探索疼痛个体差异机制，和发展克服个体差异的疼痛解码方法。第一，我们发展可精确估计疼痛诱发神经活动的差异性信息的特征提取算法，更加全面准确地描述疼痛诱发神经活动的个体差异特性。特别地，我们识别出疼痛任务引起的特异性的全脑连接模式的改变，突破了传统方法仅关注局部脑区活动的局限，证明了疼痛任务对全脑网络的调制作用(Li et al., Human Brain Mapping, 2022)。第二，我们联合使用脑电和磁共振影像探究疼痛解码个体内差异和个体间差异的机制，识别这两类个体差异的高时空分辨率脑电标记和结构功能磁共振影像标记。例如，我们发现大脑默认网络主导的动态脑连接状态的出现频率与疼痛敏感性是正相关的(Yuan et al., Behavioural Brain Research, 2019)。第三，我们基于先进机器学习算法发展疼痛解码模型，使其可以克服个体差异影响，更准确可靠地解码疼痛。例如，我们提出根据疼痛活动的相似性为每一名被试的每一个疼痛试次动态训练特定的解码模型，从而有效克服了疼痛解码中个体差异的问题(Lin et al., NeuroImage, 2021)。本项目的研究成果增强对大脑感知疼痛模式及其个体差异的理解，为推进疼痛脑电解码的临床转化提供理论基础和技术支持。