The reconstruction of objects in three dimensions, from two-dimensional MRI (Magnetic Resonance Imaging) slices is a powerful tool for better understanding of anatomical structures. In this work, a combination of image processing and computer graphics techniques has been used to develop the required algorithms for surface rendering of anatomical structures from MRI data. Surface rendering is performed in two stages. In the first stage, image processing techniques have been used to develop the required algorithms for the extraction of the data, which is then transferred to a computer graphics environment. The proposed algorithms are for successive automatic segmentation using Markov Random Fields or simple amplitude thresholding where possible, automatic starting point finding and boundary extraction, B-spline interpolation, matching and aligning the data. The last three steps, which are interpolation, matching and aligning the extracted data serves as a preparation stage for the computer graphics environment and have been developed considering this second stage. Then, the data is converted to a computer graphics data transfer format and transferred to a graphics hardwardsoftware environment. In this stage, where the success depends very much on the success of the first stage, the required computer graphics techniques such as B-spline surface/curve fitting, skinning, texture mapping, rendering, and animation are investigated for a more realistic and accurate surface rendering of anatomical structures.

Date of this Version

December 1994