基于结构检测模型的白内障连续环形撕囊边界检测方法

Boundary Extraction for Capsulorhexis of Cataract Surgery Using Structure Detection Model

针对临床白内障手术面临的连续环形撕囊术中边界自动提取的问题,提出了一种具有高计算效率和良好鲁棒性的新颖检测方案来提取撕囊边界信息;首先,应用包括一般直方图均衡化和自适应直方图均衡化在内的预处理程序来增强输入图像的对比度;然后,建立了一种结构检测模型来检测低对比度条件下细小的撕囊边界信息;在这之后,应用局部阈值和后处理来消除干扰像素;最后,经过数据集对提出的方法进行验证;实验结果表明,所提方法具有优异检测性能并取得了F测度为0.915的结果;证明了该方法可以检测出白内障手术连续环形撕囊边界,并具有较高的计算能力和精确度。

Main goal of this paper is to address various challenges towards clinical ophthalmology in the context of automatic boundary extraction in continuous curvilinear capsulorhexis of cataract surgery via structure detection model.A novel extraction scheme with high computational efficiency and robustness is proposed to extract the boundary information.Firstly,pre-processing procedure is applied that includes general histogram equalization and contrast limited adaptive histogram equalization was utilized to enhance the contrast of the surgery images.Then,a proposed structure detection model is established to detect the tiny boundary with low contrast.After that,local threshold and post process are applied to remove disturbance.Finally,the proposed method has been verified through the data sets.Experimental results show that the proposed method could extract the boundary of capsulorhexis and gets the result of F-measure=0.915,which outperform other schemes,and achieve high computational ability and accuracy.