基于多模态影像和遗传基因筛查对原发性淋巴水肿临床分期和分级精准评价的人工智能分析

Primary lymphedema is a group of lymphoid reflux disorders caused by abnormal lymphatic structure and dysfunctional development. Because of the lack of effective means of evaluation, the pathogenesis and evolution model of the disease is still not fully elucidated. Accurate evaluation of the clinical stages and classification of different types of lymphedema is the key to determine the choice of clinical treatment plan and the prognosis of the curative effect. According to previous studies on the clinical evaluation of pathogenesis of lymphedema based on multimodal imaging, This research topic is to further explore the accurate diagnosis and Quantitative assessment of the clinical typing and staging and grading of the primary lymphedema by multimodal imaging and family genetic screening and establish correlation using gene detection results with clinical phenotypes and complete genetic classification for primary lymphedema. Artificial intelligence analysis method was applied to collection and fusion of characteristic data from clinical, multimodality and genetic screening for different types of primary lymphedema. The purpose is to provide theoretical evidence for accurate diagnosis of different types of lymphedema, including the pathogenesis and clinical staging/grading system, and build standardized database, clinical diagnosis and grading path model and computer-aided intelligent diagnosis platform.

原发性淋巴水肿是一组因淋巴管结构和功能发育异常而导致的淋巴液回流障碍性疾病。因缺乏有效评估手段，其发病机制和演变模式仍未完全阐明。不同类型淋巴水肿的临床分期和分级的精准评判是决定临床治疗方案选择和疗效预后的关键。本项目在前期多模态影像学对淋巴水肿发病机制临床研究的基础上，为深入探讨其临床分型、分期及分级的精确诊断和量化评估体系，拟对原发性淋巴水肿患者行多模态影像检查和家系遗传基因筛查，利用基因检测结果和淋巴水肿临床表型建立相关性，完成原发性淋巴水肿的基因分类。采用人工智能分析方法对不同类型原发性淋巴水肿的临床、多模态影像和遗传基因筛查等多指标大数据进行特征级信息融合。旨在为评判不同类型淋巴水肿的发病机制和临床分期分级评估体系的精准诊断提供理论依据，并构建标准化的数据库、临床诊治分级路径模型和计算机辅助智能诊断平台。

原发性淋巴水肿是一组因淋巴管结构和功能发育异常而导致的淋巴液回流障碍性疾病。因 缺乏有效评估手段，其发病机制和演变模式仍未完全阐明。不同类型淋巴水肿的临床分期和分级的精准评判是决定临床治疗方案选择和疗效预后的关键。本项目的主要内容包括（1）探讨原发性淋巴水肿的临床分型、分期及分级的精确诊断和量化评估体系；（2）对原发性淋巴水肿患者行多模态影像检查和家系遗传基因筛查；（3）分析多模态影像表型、基因检测结果和临床表型的相关性，有可能完成原发性淋巴水肿的基因分类；（4）采用人工智能分析方法对不同类型原发性淋巴水肿的临床、多模态影像和遗传基因筛查等多指标大数据进行特征级信息融合，构建标准化的数据库、临床诊治分级路径模型和计算机辅助智能诊断平台。.共收集完成原发性淋巴水肿的病例2680例，其中下肢淋巴水肿1782例、小肠淋巴管扩张症292例、肺淋巴水肿病例388例和乳糜漏患者210例。同时完成了200例患者血液采样和54例的手术组织标本采样的遗传基因检测的前期工作。建立了中国第一个以淋巴水肿疾病的临床、实验室、影像和病理的大数据平台，完成构建原发性下肢淋巴水肿和肺部淋巴水肿的多模态影像AI分析平台。为淋巴水肿患者的精准定量诊断和采取科学的治疗方法以及风险因素和预后疗效的预测提供了影像学依据，同时将淋巴管道的多模态影像学检查方法推广至全国三级和二级医院，更好地为淋巴水肿患者的早期诊断、精确治疗和合理预防提供影像学支持和评判。