作者: Robert Koprowski , Renato Ambrósio
DOI: 10.1016/J.COMPBIOMED.2015.09.007
关键词: Scheimpflug principle 、 Intraocular pressure 、 Vibration 、 Computer vision 、 Keratoconus 、 Artificial intelligence 、 Computer science 、 Cornea 、 Image processing 、 Pixel 、 Linear discriminant analysis
摘要: BackgroundOne of the current methods for measuring intraocular pressure is air-puff method. A tonometer which uses this method Corvis device. With ultra-high-speed (UHS) Scheimpflug camera, it also possible to observe corneal deformation during measurement. The use modern image analysis and processing allows higher harmonics deflection above 100Hz. Method493 eyes healthy subjects 279 patients with keratoconus were used in measurements. For each eye, 140 images recorded Each was every 230?s had a resolution 200×576 pixels. new, original algorithm has been proposed. It enables separate eyeball reaction as well low-frequency high-frequency deformations from eye response an air puff. Furthermore, classification based on decision trees ResultsThe obtained results confirm possibility distinguish between subjects. features are directly related vibrations. They only available proposed software provide specificity 98%, sensitivity-85%, accuracy-92%. This confirms usefulness type that vibrations measurement tonometer. DiscussionWith new allowing separation individual image, to: automatically measure few characteristic points cornea, obtain fully repeatable same registered sequence vibration parameters large inter-individual variability patients. We showed how could be measured.We measured patients.We classification.