With optical coherence tomography (OCX) , doctors can obtain clear tomography layer structures of retina. To automatically extract the contour of the retinal sub-layers through image segmentation is a basic issue to the application of OCX in retinal diseases diagnosis. A multi-step approach, which includes the filtering, peak detection, Snake model, greedy algorithm and spline interpolation, was devised. Therefore, we realized the automatic segmentation of retinal layers and quantitative measurement of the retinal thickness. The method was successfully applied on a database of 24 images of normal people' s eyes. The result, which the automated thickness measurements derived by this algorithm was compared with thickness measurements from manually marked boundaries, indicated that this method had a good agreement with manually marked method. In conclusion, the proposed approach is promising for investigating for retinal variability studies.%利用光学相干层析(optical coherence tomography,OCT)技术可以得到清晰的视网膜层状结构,实现视网膜层状结构自动分割功能是解决OCT技术应用于视网膜疾病诊断的一项基础问题.本文通过图像平滑、峰值探测、Snake模型、贪婪算法和样条插值等综合技术,对OCT视网膜图像进行分割,自动提取层状结构轮廓并获取视网膜厚度分布图.将以上算法应用于24例正常人眼底图像,并与专家手动标记轮廓提取的厚度相比,结果证实上述视网膜自动测量算法与专家人工标记取得较好一致性.本文提出的测量算法有望应用于视网膜变异性评估.
展开▼
