Study Of Several Key Issues In The Visualization Of OCT Data

Optical Coherence Tomography (OCT) is a high-resolution biomedical imagingtechnique, which has the advantages of high-resolution, non-invasive, fast and 3Dimaging capacity. It has attracted great interests of scientists and clinicians, andbecomes a research hotspot. At present, however, there are few publicationsconcerning the signal and image processing aspects of OCT.In this thesis, the intensity of the interfered low-coherent laser (the OCT signal)is analyzed in detail at first. And then, the differences between the simplifiedtheoretical model and the actual conditions of the experiment are discussed. From theperspective of system, we propose an OCT systematic model, which illustrates theconvolution and distortion characteristic of OCT signal.Because of the intrinsic limitation of OCT, its image quality and signal-to-noiseratio are very low;pseudo coloring is often used to improve the visual appearance. Inorder to reduce the noise and enhance the images, we propose a visualizationalgorithm to improve the image quality. The algorithm includes the following aspects:background subtraction, high-frequency noise reduction, histogram equalization,color filtering etc. These methods are all efficient, so the optimized algorithms can beembedded in the real-time OCT system easily.Color transfer is a recently proposed method for colorizing grayscale images. Itcan transmit the "real" color components of a reference source color image to a targetgray image which has similar contents. Compared with the traditional pseudocoloring method, color transfer provides more objective results, and needs little userinteraction. For further improvement, swatches are used to guide and control theprocess of color transfer.In order to improve the computation efficiency of the colorizing process andsuppress the wrong hues and noise points brought in by the outliers, a patternrecognition based color transfer algorithm is proposed. It improves the quality of thecolorized images and makes the color transfer much faster. And we can then apply theimproved color transfer method to colorize the OCT gray images. The resultant colorimage retains the original luminance but has the "real" color components.