Department of Ophthalmology, Duke University Medical Center, Durham, NC 27710, USA.
Am J Ophthalmol. 2013 Aug;156(2):304-11. doi: 10.1016/j.ajo.2013.03.012. Epub 2013 May 6.
To address the misrepresentation of the eye in retinal optical coherence tomography (OCT) images and to examine the effect of this misrepresentation on retinal thickness measurements.
Prospective case series.
Five subjects with recent orbital magnetic resonance imaging (MRI) scans and normal eye examinations were consented from the clinics of the Duke Eye Center. Each subject had both eyes imaged using a retinal spectral-domain OCT system and ocular biometry measured. Two types of individualized optical models of the subject eyes-numerical and analytical-were used to determine the spatial paths of the OCT A-scans. These paths were used to reorient the A-scans in the associated retinal OCT images and generate corrected images. Using curvature as a general measure of shape, the radii of curvature of the retinal pigment epithelium in the original and corrected OCT images were compared to the ocular radii of curvature in the MRI images. Differences between the retinal thickness maps derived from the original and corrected OCT images were then determined.
The retinal curvatures were substantially flatter in the original OCT than in the MRI images (mean paired difference: 52.8 ± 41.8 mm, P < .001). Correcting the OCT images decreased the paired differences between OCT and MRI (numerical: 1.6 ± 2.3 mm, P = .091; analytical: 1.9 ± 4.3 mm, P = .278). Retinal thickness measurements between the corrected and uncorrected images differed, with a root mean square difference of 5.61 μm over the entire 6-mm extent of the image; this difference was greater peripherally (6.02 μm) than centrally (2.54 μm).
Optically based algorithms can be used to correct the shape of the retina as represented in OCT; this correction makes OCT more consistent with other clinical imaging techniques. Resultant retinal thickness maps were minimally affected by the change in shape. Ocular shape correction should be considered in future development of posterior segment OCT-based morphologic measurements.
解决视网膜光学相干断层扫描（OCT）图像中眼睛的失真问题，并研究这种失真对视网膜厚度测量的影响。
前瞻性病例系列研究。
从杜克眼科中心诊所征得 5 名最近进行过眼眶磁共振成像（MRI）扫描且眼部检查正常的受试者同意。每位受试者均使用视网膜频域 OCT 系统对双眼进行成像，并进行眼生物测量。使用两种类型的受试者眼个体化光学模型（数值和分析）来确定 OCT A 扫描的空间路径。这些路径用于重新定位相关视网膜 OCT 图像中的 A 扫描并生成校正图像。使用曲率作为形状的一般度量，比较原始和校正 OCT 图像中视网膜色素上皮的曲率半径与 MRI 图像中的眼曲率半径。然后确定从原始和校正 OCT 图像得出的视网膜厚度图之间的差异。
原始 OCT 中的视网膜曲率明显比 MRI 图像平坦（平均配对差异：52.8 ± 41.8mm，P <.001）。校正 OCT 图像减少了 OCT 和 MRI 之间的配对差异（数值：1.6 ± 2.3mm，P =.091；分析：1.9 ± 4.3mm，P =.278）。校正和未校正图像之间的视网膜厚度测量值存在差异，整个图像 6mm 范围内的均方根差异为 5.61μm；这种差异在周边处更大（6.02μm），在中心处更小（2.54μm）。
基于光学的算法可用于校正 OCT 中表示的视网膜形状；这种校正使 OCT 与其他临床成像技术更加一致。形状变化对视网膜厚度图的影响最小。在未来基于 OCT 的后节形态学测量的开发中应考虑眼球形状校正。
