University of California, Berkeley and University of California, San Francisco, Joint Graduate Group in Bioengineering, Berkeley, California 94720, USA.
J Biomed Opt. 2011 Mar;16(3):036002. doi: 10.1117/1.3548880.
Microscopic features of the human retina can be resolved noninvasively using an adaptive optics scanning laser ophthalmoscope (AOSLO). We describe an improved method to track and quantify the speed of moving objects in AOSLO videos, which is necessary for characterizing the hemodynamics of retinal capillaries. During video acquisition, the objects of interest are in constant motion relative to the background tissue (object motion). The background tissue is in constant motion relative to the AOSLO, due to continuous eye motion during video recordings (eye motion). The location at which AOSLO acquires data is also in continuous motion, since the imaging source is swept in a raster scan across the retina (raster scanning). We show that it is important to take into consideration the combination of object motion, eye motion, and raster scanning for accurate quantification of object speeds. The proposed methods performed well on both experimental AOSLO videos as well as synthetic videos generated by a virtual AOSLO. These methods improve the accuracy of methods to investigate hemodynamics using AOSLO imaging.
使用自适应光学扫描激光检眼镜 (AOSLO) 可以非侵入性地解析人视网膜的微观特征。我们描述了一种改进的方法来跟踪和量化 AOSLO 视频中移动物体的速度，这对于描述视网膜毛细血管的血液动力学是必要的。在视频采集过程中，感兴趣的物体相对于背景组织（物体运动）不断运动。由于在视频记录期间眼睛不断运动（眼睛运动），背景组织相对于 AOSLO 也在不断运动。由于成像源在视网膜上以光栅扫描的方式扫过（光栅扫描），因此获取数据的位置也在不断运动。我们表明，考虑到物体运动、眼睛运动和光栅扫描的组合对于准确量化物体速度非常重要。所提出的方法在实验 AOSLO 视频以及由虚拟 AOSLO 生成的合成视频上都表现良好。这些方法提高了使用 AOSLO 成像研究血液动力学的方法的准确性。
