Biomedical Data Interpolation for 3-D Visualization

摘要

Medical imaging devices that produce three-dimensional data usually produce thedata in the form of image slices. In such images, the resolution in z direction is lower than in x and y directions. Before extracting and displaying objects in such images, an interpolated 3-D gray-scale volume image can be generated via image interpolation techniques to fill in the missing information. The subject of this thesis is the applying three different interpolation techniques to generate intermediate slices and comparing their qualities. The three interpolation techniques are linear interpolation, cubic spline interpolation, and Fourier in terpolation. We also apply the CT image matching method, developed by Ardeshir Coshtasby, David A. Turner, and Laurens V. Ackerman, which can determine the correspondence between points in two images. Finally, we use the human visual perception model to measure the qualities of interpolation images. Linear interpolation is shown to be the best of the three interpolation techniques used in this thesis. This research also shows that without the image segmentation or the image matching process poor intermediate images will be generated.