Medical images, acquired with Computer Tomography or Magnetic Resonance Imaging are widely used for medical diagnostics. The processing and 3D reconstruction of such images offers physicians more possibilities for accurate and timely diagnosis. The goals of this work is to propose medical image processing methods that are connected with image segmentation, region selection, selected regions contour processing and 3D reconstruction, as well as to evaluate various 3D model construction methods. For region selection an algorithm, based on the existing region-growing segmentation algorithm described in [1] is proposed. For acquiring of control points a contour processing algorithm is proposed that selects control points on the contour of a given region. For 3D reconstruction voxel-based and polygonal-based models were chosen. For the creation of the polygonal model a triangulation algorithm is proposed, which combines acquired control points in triangle-shaped polygons, creating a mesh. The created voxel and polygonal models are then evaluated and compared to a model obtained in 3D imaging software 3D-Doctor. Based on the experimental results, conclusions are made about positive and negative features of each of the 3D models. Best results are obtained by combining the polygonal and voxel models, although the quality of the obtained surface is dependant on the initial voxel form. As seen from experiments, the only voxel form free of aliasing effect is the sphere. Acquired results can be used to develop a new 3D visualization method that would not have the reviewed methods negative aspects