Clinical Research Center, School of Optometry, University of California, Berkeley, CA 94720-2020, USA.
Eye Contact Lens. 2013 May;39(3):239-46. doi: 10.1097/ICL.0b013e31828dc7f3.
To improve the digital quantification of fluorescence intensity of sodium fluorescein instilled on corneal surface by modifying a slit lamp hardware and performing computerized processing of captured digital images.
The optics of a slit lamp were modified to remove corneal Purkinje reflection and to expand the illuminated area on the cornea, followed by postexperiment image processing to minimize the influence of uneven illumination. To demonstrate the feasibility and reliability of this new technique, we applied it to objective grading of corneal staining with sodium fluorescein. The results of computerized grading were compared with the results obtained using standard subjective grading of corneal staining. Objective digital grades, staining area, and staining pixel with manually and automatically defined threshold (SP-M and SP-A) were calculated for both original and processed images. Standard subjective grades of the original images were performed by 13 trained observers using National Eye Institute (NEI), Efron, and CCLRU grading scales. A series of linear regression analyses were performed to investigate the correlation between objective and subjective grades.
Digital grades of the captured images were correlated significantly with subjective grades. After minimization of the artifact caused by the nonuniform illumination, correlations between digital and subjective grading were mostly strengthened. In some cases, digital grading of corneal staining was more sensitive than subjective grading methods when differentiating subtle differences of corneal staining.
Modifications performed on commercial slit-lamp hardware and the proposed digital image-processing technique have improved the quality of captured images for semiautomated quantification of fluorescein intensity on the cornea.
通过修改裂隙灯硬件并对捕获的数字图像进行计算机处理,提高滴注在角膜表面的荧光素钠的荧光强度的数字化定量。
修改裂隙灯的光学系统以去除角膜浦肯野反射并扩大角膜上的照明区域,然后对实验后的图像进行处理以最小化不均匀照明的影响。为了证明这种新技术的可行性和可靠性,我们将其应用于角膜荧光素染色的客观分级。通过比较计算机分级与使用角膜染色标准主观分级获得的结果来证明。对原始图像和处理后的图像计算了计算机化分级、染色面积和手动和自动定义阈值的染色像素(SP-M 和 SP-A)。使用国家眼科研究所 (NEI)、Efron 和 CCLRU 分级量表,13 名经过培训的观察者对原始图像的标准主观分级进行了评估。进行了一系列线性回归分析以研究客观和主观分级之间的相关性。
捕获图像的数字分级与主观分级显著相关。在最小化非均匀照明引起的伪影后,数字和主观分级之间的相关性大多得到了加强。在某些情况下,与主观分级方法相比,角膜染色的数字分级在区分角膜染色的细微差异时更敏感。
对商业裂隙灯硬件进行的修改和提出的数字图像处理技术提高了捕获图像的质量,从而实现了对角膜上荧光素强度的半自动定量。