基于光学相干断层扫描成像的眼部三维前房角开放距离测量方法

摘要

In this article, an automatic method for measuring 3 - dimensional (3D) anterior chamber eye angle opening distance (AOD) with Fourier-domain optical coherence tomography (OCT) was proposed. A special 3D "Angle Arc" scan pattern composed of 25 cross-sectional OCT image frames was designed. The initial boundaries of the anterior and posterior cornea and anterior iris were located in all the frames and then refined. Human input was needed to identify Schwalbe's line ( SL) in 3 key frames of each 3D dataset. Next, the computer algorithm was employed to estimate SL locations in the rest 22 frames according to a mapping method. Last, the angle opening distance based on the SL ( AOD-SL) was calculated for all the frames. Result showed that the computer algorithm provided a robust and accurate method to measure 3D AOD-SL. The precise angle boundary detection was 96. 06% in image frames, and accuracy of the AOD-SL measurement in the key frames was 95. 11% by using the automated AOD-SL measurements. The automated AOD-SL measurements agreed well with the human measurements with computer calipers.%为了从光学相干断层扫描(OCT)成像的三维前房角图像中,准确的获取各连续帧中地房角开放距离(AOD),提出一种三维前房角成像模式,并针对其特性实现了房角开放距离的自动测量.该方法首先从构成一组三维图像的25帧二维房角截面图像中分割出前房角区域,并检测出角膜前后边缘和虹膜前边缘等3条初始边缘；然后基于初始边缘从原图像的梯度图中细检测得到准确的边缘；接着利用映射方法根据用户手工输入的3个关键帧图像中的Schwalbe线(SL),计算得到其余22帧原始图像中的SL位置,并最终得到每一帧图像中基于SL的房角开放距离(AOD-SL).实验结果表明,该自动测量方法具有较高的鲁棒性和准确度,其对房角边缘的检测准确率为96.06％；在关键帧中对AOD-SL测量的准确率达到95.11％.