基于大脑-小脑功能连接及机器学习方法预测大脑语言区胶质瘤患者围手术期语言功能状态及重塑机制的研究

Prior studies have demonstrated that cerebellum involved in language processing via cerebro-cerebellar network. Consistently, we found that the left cerebral gliomas involving language network areas lead to the decline of neural activity in language-related cerebellar areas and the alteration of the cerebro-cerebellar functional connectivity. Moreover, the preoperative cerebro-cerebellar functional connectivity can be used to predict the preoperative language ability at the individual subject level. We thus hypothesize that 1) the cerebro-cerebellar functional network will be remodeled after operation and such plasticity closely related to the language abilities; 2) the cerebro-cerebellar functional connectivity could be a neuro-biomarker to predict the language abilities after surgery. To test these hypotheses, we will recruit 50 patients and acquire the longitudinal follow-up data, including clinical, neuroimaging and language assessment data. Using resting-state fMRI data, we will construct the cerebro-cerebellar functional network and further delineate its relationship with changes of language abilities. We will further adopt machine-learning multivariate pattern regression techniques to predict the short-term and long-term postoperative language abilities using the cerebro-cerebellar functional connectivity. Finally, we will compare the functional connectivity that contributed the most to the prediction of short-term and long-term postoperative language abilities, which could reveal the underlying functional network mechanism of the remodeling of language after surgery. This project could potentially help establish the strategy for rehabilitation of language function in patients with gliomas involving language network areas.

研究表明，小脑通过大脑-小脑网络参与语言功能。我们前期发现，大脑语言区胶质瘤可致小脑语言功能相关区域神经功能活动水平降低和大脑-小脑功能连接改变；基于术前的大脑-小脑功能连接可个体化预测术前的语言功能评分。基于此，我们假设：大脑-小脑功能网络在此类患者手术前后可发生动态重塑，且与语言功能的动态变化密切相关；大脑-小脑网络功能连接可作为预测术后语言功能的影像学标记物。本项目拟首先对50例此类患者开展临床-影像-语言评估的纵向随访，基于静息态fMRI数据，构建手术前后的大脑-小脑网络功能连接，研究其动态变化及与语言功能变化的关系；继而采用机器学习多元模式回归方法，以术后近期大脑-小脑网络功能连接为特征，分别预测术后近期和远期的语言评分；通过比较对预测术后近期和术后远期语言评分有贡献的功能连接，揭示手术前后语言功能重塑的脑网络机制。该项目有望为语言区胶质瘤患者语言功能康复策略的制定提供科学依据。