A new algorithm and problems in automatic anterior eye chamber volume determining

This study investigates the problem of automatic determination of the anterior eye chamber volume using previously published as well as new algorithms of image analysis and processing proposed by the authors. A new method for determining the anterior eye chamber volume that provides more accurate results has been proposed. The entire algorithm was implemented in Matlab and C language. 60,000 images were obtained using OCT SS-1000 CASIA and Zeiss Visante OCT. The acquired images of the anterior segment of the eye had a resolution of 256×1024 pixels with a measuring range of 8×16 mm2. The images were acquired during routine medical examination at the Clinical Department of Ophthalmology, District Railway Hospital in Katowice, Poland, and were analysed in accordance with the Declaration of Helsinki. The new algorithm uses edge detection, morphological operations, binarization and filtration. The proposed algorithm gave the anterior chamber surface measurement error at the level of 4.3% and the anterior chamber volume measurement error of 12%. For comparison, the surface measurement error of the tomograph software was at 6.7%. Thus the obtained results were better by 2.4%. The proposed algorithm provides reproducible results automatically at a runtime of 3 s per patient using a Core i7 PC computer with 8 GB of RAM.