注意缺陷多动障碍（ADHD）是高发的儿童神经发育障碍性疾病，其社会影响大、疾病负担重。该病属于复杂性脑疾病，伴随多维度认知功能和行为缺陷，虽然大量研究发现ADHD患者存在脑电节律、多个脑区及多重脑功能网络异常，然而，由于该病的异质性强，单一指标并不能用于诊断，因此需要建立多维度的指标体系，发展客观诊断方法。本项目拟将深度学习技术用于ADHD的辅助诊断，整合EEG-MRI多模态影像特征及认知功能，建立并优化ADHD深度学习诊断模型；并结合分类结果解析并组合不同类型患者各维度的突出特征，构造更为精细的疾病分类系统，探索基于多模态影像技术对不同表型ADHD的划分标准，提取出能够显著识别不同表型患者的指标体系。预期发现潜在的生物学标志，促进ADHD神经病理机制研究的深入，形成ADHD智能辅助诊断技术，从而对不同表型ADHD的诊断评估及个性化治疗策略以及治疗靶点的选择提供更为可靠的科学依据。

Attention deficit hyperactivity disorder (ADHD) is a high incidence of neurodevelopmental disorders in children, placing a heavy burden on their families and society. The disease is a complex disease and accompanied by a variety of behavioral and cognitive deficits. Although a large number of studies have found that children with ADHD have abnormalities in EEG rhythm, multiple brain regions and brain functional networks. However, due to the heterogeneity of this disease, a single index cannot be used for diagnosis. So it is necessary to establish a multidimensional index system and develop an objective diagnosis method. This project intends to apply deep learning for the aided diagnosis of ADHD, integrate EEG-MRI neuroimaging features and cognitive functions, establish and optimize an ADHD deep learning diagnostic model; further construct a more precise disease classification system, explore the classification criteria of different phenotypes of ADHD based on multimodal imaging, and extract imaging indicators and their combinations that can significantly identify different phenotypes of children with ADHD. It is expected to identify potential biomarkers, contribute to the understanding of neuropathological mechanisms of ADHD, and form the aided diagnosis technology of ADHD, so as to provide more reliable scientific evidences for the diagnosis and evaluation of different phenotypes of ADHD as well as personalized treatment strategies and treatment target selection.