The Study Of Three-Dimensional Reconstruction In Medical Images

Three-dimensional reconstruction of medical images is one of the focuses ofthe current study, and it is also an inter-disciplinary field. The three-dimensionalreconstruction techniques identify the contours of a series of two-dimensionalmedical images to reconstruct three-dimensional images of human’s organs usingcomputer technologies and graphic image processing techniques. Doctors canrotate, zoom and move the3-D images to intuitively and easily observe theinternal structure and the surface of the tissue even the size and shape of thetissue. With the help of this information, doctors can make an accurate diagnosis.This paper introduces the concepts and processes of the3-D reconstructionfirstly. Imaging principle of three-dimensional reconstruction in medical imaging,data acquisition, data pre-processing techniques are detailed to illustrate theprocesses of reconstruction. The3-D interpolation is one of the difficulty in3-Dreconstruction and also is the emphasis of this paper.Interpolation is a necessary processing step in3-D reconstruction because ofthe non-uniform resolution. The three-dimensional interpolation techniques canbe divided into two categories: scene-based interpolation and object-basedinterpolation. This paper reviews and summarizes the existed interpolation andpresents a new3-D binary interpolation algorithm. Most of the conventionalinterpolation methods merely use two slices to obtain the missing slices betweenthe two slices.When the key slice is missing, those methods may fail to recover itonly employing the local information. And the surface of3-D object especiallyfor the medical tissues may be highly complicated, so a single interpolation canhardly get high-quality3D image. We propose a novel binary shape-based3Dinterpolation method. The proposed method takes advantages of the globalinformation. And this method chooses the best curve adaptively from lots ofcurves based on the complexity of the surface of3D object. The results of thismethod are compared with other interpolation methods on artificial objects andreal breast cancer tumor to demonstrate the excellent performance.Finally, a3-D interpolation comparison platform on surface curve and a3-Dreconstruction of medical images are built. The former is realized by MATLAB GUI and can be used to analyze the performance of the interpolation. The surfacerendering and volume rendering on artificial objects and real ultrasound imagesare displayed in this paper with the support of VTK and QT on the3-Dreconstruction platform. And the result of the rendering is saved for the intentionof feature extraction on3-D objects.