Study On Reconstruction Of Medical Slice Image And Optical Propagation In Biological Tissue

The reconstruction of 3-Dimensional medical images is becoming a hot topic inrecent years. To introduce 3-dimensional reconstruction into radioactive treatment,combining isodose estimation by way of Monte Carlo, is a diagnose tool with wideapplications. 3-D reconstruction and visualization of medical images can easilydisplay the physical attribute and spatial position of body tissues. It is able to providea vivid 3-D body structure image and help making a optimum treatment plan. Thedissertation builds up a platform of the radioactive treatment and simulating operation,in which the related medical research projects can be performed. The basis of reconstruction of 3D medical images is that all the imageinformation must be in an united coordinate, which has been established in this paper.In the united coordinate the image must be registered and match each other. The extraction of tissues' contour is the fundamental element in surfacereconstruction. To extract the tissue contour semi-automatically and to match thecontour with the image grads information perfectly are studied. The huge contourdata is a great challenge to the real time operation. A contour sampling algorithm withvariable resolution, based on threshold and arc distance theory, is presented. Morecontour information is gained through fewer points. The contour was adaptivelysampled and stored in matrix. The primary questions in image reconstruction, are discussed, which are thecontour match, branch and contour connection. The contour connection isemphatically discussed. The Optimal Similarity Function, which combined withwavelet theory, is constructed to extract the contour matched point semi-automatically. The new approaches for both contour match and branch are presented.A high-quality, vivid 3-D model is reconstructed. Ordinary medical image is often provide the horizontal slice image. The coronaland sagittal information are missed in the primal medical image. Therefore the primalmedical image cannot provide all the 2-dimensional information. The existing imageinterpolation methods, the intensity interpolation and object interpolation, areanalyzed. A new algorithm, which maintains boundary partly, is applied to get thecoronal and sagittal medical images. The layer interpolation of horizontal image isresolved by same algorithm. The gain of isodose curve is the final aim of radioactive treatment. Therelationship between ray and tissue is discussed. The Monte Carlo mathematic modelis established to conclude and evaluate the isodose curve. The factors may affect dosedistribution are analyzed. Multi-layered model of skin is presented to study the lightpropagation and distribution in skin tissue by using Monte Carlo technique. Therelationship between absorption energy and thickness is simulated. A new II天津大学博士学位论文Chain-Code algorithm based on eight-direction tracing is presented. Through thistracing principle, all contours' information in the chain-code are obtained finally. Theproblem of Isodose curves with same dose appearing several times in medical imageprocess is solved by this tracing principle.