3D Reconstruction Of The Macula Lutea Based On Optical Coherence Tomography Images

The modeling and simulation of virtual organ is one of the frontier research areas. As one of the most important organs of human, eye should be modeled and simulated factitiously. Fundus is one of the most important parts in eyeball, especial macula lutea, whose characteristics are of great significance to diagnosis and cure of eyes'diseases. OCT image is taken from the data of live human eyes, so the macula 3D Model based on these OCT images lays a great important role in applying dummy eyes practically, and also is the primary research for 3D OCT fundus diagnosis technology.According to the characteristics of OCT macula images, we studied some algorithms for segmentation of OCT macula images, and abstracted the region of interest from the images. Then, image sequence was registered and an interpolation algorithm was designed to get intermediate images. Finally, a 3D visualization human macula model was reconstructed. In the end,a 3D visualization demo system was developed based on personal computer and Windows operation system. This demo system provided a graphics user interface to rotate, scale the 3D objects.In this thesis, some innovations had been done in the following aspects:1,A segmentation method of mathematic morphology combining with grads was presented, which was used to segmentation the regions of interest from the image sequence.2,According to the characteristics of OCT macula images, a method based on luminance and contrast was presented to adjust the bright difference between images, and the availability was approved.3,OCT macula images were registered in a method based on maximization of mutual information. For efficiency purpose, an coarse registration using minimum mean square error standard was applied first, and Powell, a optimization algorithms,had been applied to register the images.4,An interpolation arithmetic based on linear interpolation was designed to get intermediate images; a 3D macula model was reconstructed based on these intermediate images and the origin images registered.