Dexi Zhu1, Yilei Shao, Lin Leng, Zhe Xu, Jianhua Wang, Fan Lu, Meixiao Shen. 1. School of Optometry and Ophthalmology (D.Z., Y.S., L.L., Z.X., F.L., M.S.), Wenzhou Medical College, Wenzhou, Zhejiang, China; and Bascom Palmer Eye Institute (J.W.), Miller School of Medicine, University of Miami, Miami, FL.
Abstract
OBJECTIVE: To test accuracy and repeatability of a software algorithm that performs automatic biometry of the anterior segment of the human eye imaged with long scan depth optical coherence tomography (OCT). METHODS: The ocular anterior segment imaging was performed with custom-built long scan depth OCT. An automatic software algorithm including boundary segmentation, image registration, and optical correction was developed for fast and reliable biometric measurements based on the OCT images. The boundary segmentation algorithm mainly used the gradient information of images and applied the shortest path search based on the dynamic programming to optimize the edge finding. The automatic algorithm was validated by comparison of the biometric dimensions between automatic and manual measurements and repeatability study. RESULTS: Biometric dimensions of the anterior segment, including central corneal thickness, anterior chamber depth, pupil diameter, crystalline lens thickness, and radii of curvature of the anterior and posterior surfaces of lens, were obtained by the automatic algorithm successfully. There were no significant differences between the automatic and manual measurements for all biometric dimensions. The intraclass correlation coefficients (ICC) of agreement between automatic and manual measurements ranged from 0.85 to 0.98. The coefficients of repeatability and ICC for all automatic dimensions were satisfactory (1.1%-6.1% and 0.663-0.990, respectively). CONCLUSIONS: The high accuracy, good repeatability, and fast execution speed for automatic measurement of the anterior segment dimensions on the OCT images were demonstrated. The application of this automatic biometry is promising for investigating dynamic changes of human anterior segment during accommodation in real time.
OBJECTIVE: To test accuracy and repeatability of a software algorithm that performs automatic biometry of the anterior segment of the human eye imaged with long scan depth optical coherence tomography (OCT). METHODS: The ocular anterior segment imaging was performed with custom-built long scan depth OCT. An automatic software algorithm including boundary segmentation, image registration, and optical correction was developed for fast and reliable biometric measurements based on the OCT images. The boundary segmentation algorithm mainly used the gradient information of images and applied the shortest path search based on the dynamic programming to optimize the edge finding. The automatic algorithm was validated by comparison of the biometric dimensions between automatic and manual measurements and repeatability study. RESULTS: Biometric dimensions of the anterior segment, including central corneal thickness, anterior chamber depth, pupil diameter, crystalline lens thickness, and radii of curvature of the anterior and posterior surfaces of lens, were obtained by the automatic algorithm successfully. There were no significant differences between the automatic and manual measurements for all biometric dimensions. The intraclass correlation coefficients (ICC) of agreement between automatic and manual measurements ranged from 0.85 to 0.98. The coefficients of repeatability and ICC for all automatic dimensions were satisfactory (1.1%-6.1% and 0.663-0.990, respectively). CONCLUSIONS: The high accuracy, good repeatability, and fast execution speed for automatic measurement of the anterior segment dimensions on the OCT images were demonstrated. The application of this automatic biometry is promising for investigating dynamic changes of human anterior segment during accommodation in real time.