基于医疗影像深度学习的HIV脑认知模型及其临床应用

HIV相关神经认知障碍综合征（HAND）严重损害患者认知功能，给家庭和社会带来严重负担。研究HAND临床前期无症状神经认知损伤（ANI）脑机制，对理解其发病机制和研发早诊早治新方法具有重要科学意义和临床价值。申请人基于前期研究，拟采用Frascati标准综合运用神经心理学量表筛选HAND临床前期患者，采用神经成像技术纵向随访观察抗逆转录病毒治疗前后不同时相点神经认知损伤相关评价指标，结合主观认知评估和客观神经环路结构、功能评价指标，系统研究以额叶、胼胝体控制系统为主要调控机制的ANI病理认知损伤机制，建立ANI病理损伤数学模型；进而对ANI人群治疗后随访数据脑网络动态变化规律研究，以明确ANI转归神经机制。基于上述研究结果及多模态数据训练跨维度ANI深度学习模型，通过累积数据和优化算法，逐渐增强模型临床诊断性评估效果，建立ANI多维度客观预测神经成像多模数据模型，以期用于HAND早期诊断。

HIV-related neurocognitive disorder syndrome (HAND) severely impairs the cognitive function of patients, putting a heavy burden on families and society. It is of vital importance to study the brain mechanism of pre-clinical asymptomatic neurocognitive impairment (ANI) with scientific significance and clinical value for understanding its pathogenesis and developing new methods for early diagnosis and treatment. The neurological imaging technique was used to evaluate neurocognitive impairment at different time points before and after antiretroviral therapy. Combing the cognitive assessment, objective neural function structure and function evaluation index, we study the pathological mechanism of ANI cognitive damage with the frontal lobe and corpus callosum control system as the main control mechanism. In order to establish a mathematical model of ANI pathological damage, we follow up the patients after treatment. The study is meant to determine the neural mechanism of ANI transformation and the dynamic changes of brain networks. Based on the above research results and multimodal data training cross-dimensional ANI deep learning model, we gradually enhance the clinical diagnostic evaluation and establish an ANI multi-dimensional objective prediction neural imaging multi-modal data model by the accumulation of data and optimization of algorithms, for the application of HAND early diagnosis.