Vaishnav Yash J, Rucker Stuart A, Saharia Keshav, McNamara Nancy A
.
Biomed Tech (Berl). 2017 Nov 27;62(6):609-613. doi: 10.1515/bmt-2016-0148.
Corneal confocal microscopy (CCM) is an in vivo technique used to study corneal nerve morphology. The largest proportion of nerves innervating the cornea lie within the subbasal nerve plexus, where their morphology is altered by refractive surgery, diabetes and dry eye. The main limitations to clinical use of CCM as a diagnostic tool are the small field of view of CCM images and the lengthy time needed to quantify nerves in collected images. Here, we present a novel, rapid, fully automated technique to mosaic individual CCM images into wide-field maps of corneal nerves. We implemented an OpenCV image stitcher that accounts for corneal deformation and uses feature detection to stitch CCM images into a montage. The method takes 3-5 min to process and stitch 40-100 frames on an Amazon EC2 Micro instance. The speed, automation and ease of use conferred by this technique is the first step toward point of care evaluation of wide-field subbasal plexus (SBP) maps in a clinical setting.
角膜共焦显微镜检查(CCM)是一种用于研究角膜神经形态的体内技术。支配角膜的神经大部分位于基底膜下神经丛,其形态会因屈光手术、糖尿病和干眼症而改变。CCM作为诊断工具在临床应用中的主要局限性在于CCM图像的视野小,以及量化采集图像中的神经所需的时间长。在此,我们提出了一种新颖、快速、全自动的技术,可将单个CCM图像拼接成角膜神经的宽视野图。我们实现了一个OpenCV图像拼接器,该拼接器考虑了角膜变形,并使用特征检测将CCM图像拼接成一个蒙太奇画面。在亚马逊EC2微型实例上,该方法处理并拼接40 - 100帧图像需要3 - 5分钟。这项技术所带来的速度、自动化程度和易用性是在临床环境中对宽视野基底膜下神经丛(SBP)图进行即时护理评估的第一步。