A medical image processing apparatus of the present invention includes an edge extracting section that extracts edges of an inputted two-dimensional image, a three-dimensional-model estimating section that estimates a three-dimensional model on the basis of the two-dimensional image, a voxel extracting section that extracts, on the basis of positions of respective voxels, where the edges are present, a predetermined voxel group to be set as a calculation object of a shape feature value, a shape-feature-value calculating section that calculates the shape feature value for at least a part of voxels among the predetermined voxel group, a three-dimensional-shape extracting section that extracts a voxel group, a three-dimensional model of which is estimated as a predetermined shape, on the basis of the shape feature value, and a tuberal-shape detecting section that detects the voxel group as a voxel group forming a tuberal shape in the three-dimensional model of the living tissue.