基于多模态医学影像技术的湿性年龄相关性黄斑变性诊断与分析

Wet age-related macular degeneration (AMD) is a common cause of blindness disease, and is the main form of AMD in China. The recent ophthalmic imaging techniques, such as fundus flourescein angiography and optical coherence tomography etc, provide a great help for the diagnosis and treatment of AMD. However, most of the ophthalmologists are manually measured various parameters of the retina, which is subjective and difficult to ensure accuracy and consistency, and especially difficult for 3D data analysis. This research is to develop and validate a multi-modality retinal imaging based automatic wet-AMD computed aided diagnosis system, which can quantitatively analyze various lesions ingredients in wet-AMD and identify choroidal neovascularization(CNV), polypoidal choroidal vasculopathy (PCV) and retinal angiomatous proliferation (RAP), and evaluate the consistency between the proposed system and other diagnosis techniques, and evaluate the association between the analysis results and treatment effects. This study is an interdisciplinary research, which tightly integrates the ophthalmic clinical practice and computer engineering disciplinary. The 3D automatic image registration, image segmentation, and 3D visualization techniques in computer engineering are fully applied to clinical wet-AMD diagnosis. Our objective is to improve the existing diagnosis and analysis level for wet-AMD disease, and to explore new methods and technologies.

年龄相关性黄斑变性（Age-related Macular Degeneration, AMD）是常见的致盲眼病，在我国以湿性AMD为主。眼科影像学技术，例如眼底荧光血管造影和光学相干断层扫描等，为临床眼科医生诊断和治疗湿性AMD提供了帮助。然而目前大部分眼科医生采用手动的方式对视网膜的各种参数进行分析和测量，主观性强，无法保证准确性和一致性，而且难以全面分析三维扫描带来的大量数据。本课题拟采用图像分割，校准、三维可视化等图像分析技术，对临床采集到的湿性AMD多模态视网膜影像数据进行处理与分析，建立一套自动化湿性AMD诊断和分析系统，自动识别和定量测量湿性AMD中的各种病变成分，鉴别脉络膜新生血管、息肉样脉络膜血管病变和视网膜血管瘤样增生，并评价该系统与湿性AMD的其他诊断技术的一致性，以及分析结果与治疗反应的关联。本课题为提高现有的湿性AMD疾病诊断和分析水平，探索新方法和新技术。

本研究将眼科临床医学实践与计算机工程的相结合，进行跨领域的合作，采用计算机工程学图像校准、分割建模和三维可视化等技术，对眼科多模态影像数据，如3D OCT 和眼底彩照等，开发出基于多模态视网膜医学影像的诊断和分析系统，并结合临床研究对该进行优化和验证。主要研究内容包括：（1）针对视网膜的各种病变，建立了基于多模态医学影像的自动、三维定量分析方法；（2））建立了基于多模态医学影像疾病数据的分割与分析方法；（3）研究了眼科多模态影像数据，如3D OCT 和眼底成像等的校准算法。取得重要科研成果包括：实现了基于OCT的视网膜分层算法，可对正常和病变视网膜都能实现精确分层；实现了基于眼底彩照、OCT等的视网膜病变结构的自动检测、分割和量化分析算法，可对年龄相关性黄斑变性疾病相关的多种眼底疾病，如微血管瘤、分支动脉阻塞、黄斑裂孔、囊样水肿等进行精确分析；实现了视网膜的二维眼底图像到三学相干断层的校准；已搭建完成跨平台的眼科多模态医学影像处理与分析软件系统；发表国际顶级期刊/会议论文33篇；申请中国发明专利14项（2项已授权）。本项目取得的科研成果为眼科湿性年龄相关性黄斑变性疾病的诊断与分析提供了系列高效的新技术和新方法。