糖尿病大脑神经-血管失偶联介导认知损害机制的多模态MRI研究

Diabetes (mainly type II diabetes) is one of the important risk factors for mild cognitive impairment (MCI) and dementia. The complex brain damage mechanism increases the risk of cognitive impairment, and accelerates the deterioration of cognitive function at the same time. Previous studies have shown that neurovascular uncoupling may be the major mechanism of diabetic cognitive impairment. However, there still lacks a neurovascular uncoupling network system for diabetic MCI. Advanced magnetic resonance imaging technology can explore the mismatch of cerebrovascular blood flow and nerve integrity to reflect the characteristics of neurovascular uncoupling in diabetic MCI. We previous observed that execution and memory dimension are mainly affected in diabetic MCI group. Besides, the correlation with cognitive score of characteristics of neurovascular uncoupling is higher than that of functional brain network connectivities. Therefore, we intend to integrate psychological measurement and clinical biochemical examination, and to establish the diabetic neurovascular uncoupling system based on the multimodal MRI technique, and to establish the deep learning model for individual diagnosis. The reliability and validity of this system will be tested in another diabetes group. The neurovascular uncoupling system will likely assist in screening patients with diabetic MCI, to perform early intervention so as to prevent or delay the progression from diabetic MCI to dementia.

糖尿病（II型糖尿病为主）是轻度认知功能障碍（MCI）和痴呆的重要风险因素。其复杂的大脑损伤机制既增加了认知损害的比例，又加速了认知功能的恶化。以往研究表明神经血管失偶联可能是糖尿病伴MCI患者大脑功能异常的的神经机制，但目前尚缺乏糖尿病患者神经血管失偶联网络体系。磁共振影像学技术可以通过探索大脑微血管血流与神经完整性的不匹配反映糖尿病MCI患者大脑的神经血管失偶联特性。本课题组前期观察到糖尿病伴MCI患者主要受损的认知维度为执行能力和记忆力，神经血管失偶联网络特征与认知评分的相关性较大脑功能网络连接性更高。因此，我们拟综合心理行为学和临床生化等指标,利用多模态MRI技术建立糖尿病神经血管失偶联网络体系，并建立个体化诊断的深度学习模型。随后在糖尿病人群中测试该模型的灵敏度和特异度。该模型将有助于筛选糖尿病伴MCI患者，及早干预，对预防和延缓其向痴呆发展具有重要的理论和临床实践意义。

2型糖尿病（T2DM）是轻度认知功能障碍（MCI）和痴呆的重要风险因素。其复杂的大脑损伤机制既增加了认知损害的比例，又加速了认知功能的恶化。以往研究表明神经血管失偶联可能是糖尿病伴MCI患者大脑功能异常的的神经机制，但目前尚缺乏糖尿病患者神经血管失偶联网络体系的研究。我们综合心理行为学和临床生化等指标，利用多模态MRI技术，在以往研究的基础上开发了应用影像手段探索大脑神经血管偶联特征的方法，并研发了相应的分析软件MICA。同时应用多模态MRI技术探索了大脑不同反应神经活动和脑血流指标的变化特点，建立糖尿病神经血管失偶联网络体系，证明了神经血管偶联在认知损伤中贡献大于单纯的网络连接、脑血流供应、白质纤维完整性等指标。通过随访T2DM患者的认知水平的动态变化情况，建立认知水平的神经血管偶联预测模型。证明了网络神经血管失偶联是糖尿病伴MCI患者大脑功能异常的的关键神经机制，对后续筛选发生MCI的T2DM患者、预防和延缓其向痴呆发展具有重要的理论和临床实践意义。