Visualization of biological objects involves a set of methods of registration, processing, and interpretation of images of anatomical structure of internal organs. It plays the most important role among the diagnostic methods used in biology and medicine. With the advent of high-power computer facilities, new possibilities arose in the development of this field of science. Modern methods of biological object image processing call for the application of complex computing methods and special algorithms of solving problems of visual diagnostics [1, 2] reflecting special features of scientific research and professional practice in this field. The most essential restriction that hinders wide application of modern visualization methods in clinical and research practice is a very large volume of computations necessary for obtaining informative images. The problem of insufficient computing resources is especially acute in three-dimensional image reconstruction most interesting for clinical application. There are three typical problems of image processing for visualization of biological objects: – improvement of the image quality (first of all, spatial resolution) with invariable parameters of registration equipment, – reconstruction of a three-dimensional biological object (in the region of interest) from flat cuts and generation of its mathematical model in the form of a set of finite elements, – search for an analogous image with the well-known pathology or anatomical features in a database. These problems can be solved exclusively by numerical methods because of the complex surface of most biological objects and nonuniform distribution of the quantity being visualized. The reconstruction problem can be solved by different (Newton, Newton–Gauss, conjugated gradient, genetic algorithm, neural network in parallel realization [3], etc.) optimization methods. Irrespective of the applied method, multiple solution of a series of problems on the surface approximation is required. The initial data for a solution of the problem of three-dimensional image reconstruction involve a set of flat cuts (tomographic images) obtained by one of the methods (computer, magneto-resonant, or ultrasonic tomography, radioisotope imaging, etc.) and an approximate mathematical model of the examined region refined (identified) for a concrete object. As a result of application of the finite element method, the initial problem is reduced to a system of linear algebraic equations with a positively determined symmetric matrix. This system can have very large dimensions, especially when reconstructing not only the object surface, but also its internal structure with high resolution. The typical dimensions of the model necessary for reconstruction of individual organs comprise from 7 to 10 million elements; iterative methods of solving of a large system of linear equations with several thousand iterations are used for high-quality reconstruction; as a result, the reconstruction time can be very large or the problem cannot be solved even on high-power personal computers.