Segmentace biomedicínských obrazových dat na základě aktivních kontur

Abstract

This bachelor thesis deals with the topic of current methods for automatic segmentation of retinal lesions and optical disk. BP includes an introduction to image processing, types of image formats, research on segmentation based on active contours from the image. In the practical part, the work deals with the posterior segment of the eye, RetCam3 imaging, physiological and pathological findings in premature infants, image preprocessing, implementation of active contours, segmentation of retinal images, methods used and calculation of retinal lesion and optical disc parameters. The algorithm was tested on selected 33 images from eleven patients from one database of images taken on the RetCam3 device. The outputs of segmentation are written in the conclusion. Gaussian noise completely degraded the image to such an extent that subsequent segmentation of retinal lesions is not possible.