Volume Quantification and Visualization for Spinal Bone Cement Injection

of thesis entitled “Volume Quantification and Visualization for Spinal Bone Cement Injection” submitted by Xie Kai for the degree of Master of Philosophy at the University of Hong Kong in August 2002 Imaging techniques, like X-ray, Magnetic Resonance Image (MRI) and Computer Tomography (CT), are common place in medicine nowadays. These instruments make it easier for surgeons and doctors to diagnose diseases and increase the ratio of success in surgery, as they enable internal human structures to be observed. Computer graphics, one of the branches of Computer Science, is commonly used in medical disciplines. It reconstructs the 2-dimension image slices obtained by CT or MRI to 3-dimensionel models, making it easier for surgeons and doctors to observe internal organs. Applications of computer graphics can not only visualize medical data, but also quantify and evaluate it specified usages. A joint research project was conducted with medical specialists of the Department of Orthopaedic Surgery at the University of Hong Kong. A material called bone cement is often used with fracture patients in case where traditional medical methods do not provide satisfactory results, especially in the case of spinal fractures. Doctors and surgeons are very interested in the qualification of bone cement injection, but they must depend on their experience to distinguish bone cement from other tissues in 2-dimention CT slices. There is therefore a need for the development of a computational method which classifies the bone cement automatically. However, the complicated shape and the very similar densities of bone cement to parts of other tissues makes such a task very difficult. A comparison between the sample before the injection and after the injection of bone cement is needed in order to collect the characteristics of the shape of the bone cement in the spine. In this thesis, a series of methods were developed to compare two such samples. Two major steps were involved in this procedure: Feature point detection and Point alignment. In the first step, groups of feature points were detected in both volume data sets by a statistical method which differs from the traditional feature point detection method. The second step aimed at aligning two point sets in 3-dimensionel space approximately. A method requiring only a few points among the point sets to be matched was designed to save as much running time as possible without significant loss of accuracy. Finally, two volume data sets were aligned by the matrix obtained by the second step before the comparison. Thus enables the shape of the bone cement to be easily classified by comparing the volumes voxel by voxel. The result can be more easily evaluated by the surgeons.