信息逐步更新中。。。
研究方向
1、Registration
Fundus examination of the newborn needs to be treated seriously so as to avoid blindness. As at least five images are captured from different field of view (FOV) for each eye of newborns, ophthalmologists have to review these images one by one which is a time-consuming and fallible task. This paper proposed a novel image mosaic method characterized by avoiding the useless detection and extraction of the feature points in the non-overlapping region of the paired images during registration process. First, this method utilized phase correlation algorithm to determine the overlap area of the paired images and limited the detection and extraction range of feature points. Second, the improved Speeded Up Robust Features (SURF) algorithm based on the Shannon Entropy removing zero entropy feature points was applied to accomplish registration of sense images. Finally, a weighted fusion method based on distance matrix was introduced to fuse five to seven registered images in each eye.
2、视网膜血管的分析
眼球底部的视网膜血管是全身血管系统中唯一可以无创直接在活体上观测到的微循环系统,血管形态的变化，如血管宽度、角度、分支形态等，都可以作为与血管相关疾病的诊断依据。眼科致盲疾病, 例如白内障、青光眼、糖尿病视网膜病变、老年性黄斑病变等, 甚至能直接从眼底视网膜血管病变中观察到。同时，视网膜血管也为研究神经系统疾病提供了一个窗口，视网膜循环系统与脑部循环系统有着相同的解剖生理特征，而脑血管疾病主要是以小血管病变为基础的 一类疾病。因此，观察视网膜血管的改变为脑血管疾病的早期诊断、甚至预测其危险性提供了理论依据。
3、基于深度学习的眼底影像研究
视网膜病灶分析，主要包括红色病灶、黄色病灶、血管异常。红色病灶包括微血管瘤，点状、片状、线状出血等，黄色病灶分为硬性渗出物，软性渗出物，棉绒斑等，血管异常分为新生血管，视网膜毛细血管异常，静脉环，静脉串珠等。识别出的视网膜眼底病灶能够确诊一些疾病，包括眼底疾病和全身性疾病。同时，病灶数量、面积、所在区域等为疾病的分期或分级提供了依据。
添加评论