基于SD-OCT图像的中浆病变区域分割
本文针对现有的26组中浆病变的OCT图像进行自动分割，首先采用三文图搜索的方法进行层分割，该方法能准确分割正常眼睛的各层边界，对于病变较小的中浆图像也有较好的鲁棒性，但对于病变区域较大的CSC图像，分割方法还需进一步改进，接着在限制范围内通过自适应阈值及区域生长法准确找到中浆病变区域，并将其分割出来。实验结果表明：本文算法能够准确分割出CSC边界，且适用于三种形态下的中浆病变。
毕业论文关键词 中浆 频域光学相干层析图像 三文图搜索算法 层分割 自适应阈值 区域生长
毕业设计说明书外文摘要
Title CSC Region Segmentation Based on SD-OCT Images
Abstract
Central serous chorioretinopathy (CSC), is a common macular disease. Spectral Domain Optical Coherence Tomography (SD-OCT) for short retinal images can clearly show the micro structure of the posterior segment of eyes, so many characteristic of CSC lesion area has been clearly shown in OCT images, which makes the diagnosis of CSC more clearly and plays a very important role in the study of disease.
In this paper, we segment twenty-six cubes of OCT images of CSC automatically. Firstly, we segment the layers in OCT retinal images through 3-D graph search methods. It can accurately segment all the layers of normal eyes, and it is robust to images with small lesion area. But for images with large lesion area, the method still needs to be improved. Then we segment the area accurately within the range of restriction through adaptive threshold and region growing. The experiment results indicate that the algorithm can segment the boundaries of CSC lesion area accurately, and it is useful for three forms of CSC.
Keywords CSC SD-OCT 3D graph search algorithm layer segmentation
adaptive threshold region growing
目 次
1 绪论 1
1.1 光学相干层析成像背景介绍 1
1.2 视网膜SD-OCT图像分层结构介绍 1
1.3 视网膜层分割的研究现状 2
1.4 中浆病变相关介绍 2
1.5 中浆病变的图像特征及研究意义 2
1.6 论文的组织结构 3
2 视网膜图像层分割 4
2.1 图像预处理 4
2.2 最优三文图搜索算法理论基础 5
2.2.1 三文图结构的构建 6
2.2.2 三文图结构顶点的权重设计 8
2.3 实验结果及分析 9
2.3.1 正常眼睛的视网膜层分割结果分析 10
2.3.2 CSC图像的视网膜层分割结果分析 10
2.4 本章小结 11
3 基于自适应阈值和区域生长的CSC分割 12
3.1 基于自适应阈值的图像分割 12
3.1.1 算法理论基础 12
3.1.2 算法实现过程 14
3.2 区域生长算法 14
3.2.1 基于人类视觉模型的相似性准则 15