Surface modeling--uncertainty estimation and visualization.

The work presents a method of determining uncertainty of surface-based models that are reconstructed fragments of a human body built on the basis of slice images received from computed tomography CT or magnetic resonance imaging MRI. An analysis of geometric structure of the models has been carried out determining features, such as local inclination angle of the normal to the surface relative to the direction of scanning, and local radius of curvature. These features, together with the distance between slice images, have the largest influence on accuracy of the reconstructed surface-based models. A model of uncertainty has been determined by comparing properly selected virtual anatomical models and their spatial reconstructions. The estimated uncertainty model has then been employed to determine local errors in geometric structure of surface-based models. A quality visualisation of the errors in the geometric model has been presented in the form of a colour scale, and a quantity visualisation in the form of a ribbon, whose width is proportional to the uncertainty model.