Shadow Removal and Contrast Enhancement in Optical Coherence Tomography Images of the Human Optic Nerve Head
Shadow Removal and Contrast Enhancement in Optical Coherence Tomography Images of the Human Optic Nerve Head
复制标题
DOI：
10.1167/iovs.10-6925
复制
发表时间：
2011-09-01
影响因子：
4.4
通讯作者：
Mari, Jean Martial
中科院分区：
文献类型：
--
作者：
Girard, Michael J. A.;Strouthidis, Nicholas G.;Mari, Jean Martial
目的.目的：提高光学相干断层扫描（OCT）对视神经乳头（ONH）的成像质量。开发了两种算法，一种用于补偿光衰减，另一种用于增强OCT图像的对比度。前者是借用超声成像的发展，并被证明适用于时间或频谱域OCT。后者是基于像素强度指数的直接应用。这两种算法的性能进行了测试的频谱域OCT图像的四个成人ONHs. patients。补偿算法的应用显着降低了层内对比度（从0.74 +/- 0.16降至0.17 +/- 0.12; P < 0.001），表明成功去除了血管阴影。此外，补偿显著改善了更深ONH组织的可见性，例如视乳头周围巩膜和筛板。对比度增强算法的应用显著增加了层间对比度（从0.48 +/- 0.22增加到最大值0.89 +/- 0.05; P < 0.001），因此可以更好地区分组织边界。只有在考虑补偿的情况下，对比增强才是可靠的。所提出的算法是简单的，可以显着提高质量的ONH图像临床捕获OCT。这项研究具有重要意义，因为它将有助于提高我们的能力，执行自动分割的ONH;量化的形态和生物力学的ONH组织在体内;并确定潜在的风险指标青光眼。（Invest Ophthalmol维斯科学。2011;52：7738-7748）DOI：10.1167/iovs.10-6925
PURPOSE. To improve the quality of optical coherence tomography (OCT) images of the optic nerve head (ONH).METHODS. Two algorithms were developed, one to compensate for light attenuation and the other to enhance contrast in OCT images. The former was borrowed from developments in ultrasound imaging and was proven suitable with either time- or spectral-domain OCT. The latter was based on direct application of pixel intensity exponentiation. The performances of these two algorithms were tested on spectral-domain OCT images of four adult ONHs.RESULTS. Application of the compensation algorithm significantly reduced the intralayer contrast (from 0.74 +/- 0.16 to 0.17 +/- 0.12; P < 0.001), indicating successful blood vessel shadow removal. Furthermore, compensation dramatically improved the visibility of deeper ONH tissues, such as the peripapillary sclera and lamina cribrosa. Application of the contrast-enhancement algorithm significantly increased the interlayer contrast (from 0.48 +/- 0.22 to a maximum of 0.89 +/- 0.05; P < 0.001) and thus allowed a better differentiation of tissue boundaries. Contrast enhancement was robust only when compensation was considered.CONCLUSIONS. The proposed algorithms are simple and can significantly improve the quality of ONH images clinically captured with OCT. This study has important implications, as it will help improve our ability to perform automated segmentation of the ONH; quantify the morphometry and biomechanics of ONH tissues in vivo; and identify potential risk indicators for glaucoma. (Invest Ophthalmol Vis Sci. 2011;52:7738-7748) DOI:10.1167/iovs.10-6925