Image processing apparatus and image processing method

A method of controlling an image processing apparatus for extracting a predetermined shape with respect to a medical image, the method performing:

• a first feature value calculation step of calculating, by a first feature value calculation unit, a first feature value for each pixel in an image picked up of living mucosal structure based on eigenvalues obtained through computations using a Hesse matrix, where the first feature value represents a value of an index which indicates what shape a local region including a pixel of interest and each pixel in a neighborhood of the pixel of interest has;
• a second feature value calculation step of calculating, by a second feature value calculation unit, a degree of concentration of a gradient vector as a second feature value based on the gradient vector defined by a gradient direction and a gradient intensity calculated for each pixel in the image, where the degree of concentration of the gradient vector represents a state of distribution in the local region;
• an evaluation value calculation step of calculating, by an evaluation value calculation unit, a geometric evaluation value for each pixel in the image, where a value which enables distinguishing between a linear mucosal structure and a massive mucosal structure contained in the image and which does not depend on contrast of the image is calculated as the geometric evaluation value based on calculation results of the first feature value and the second feature value;
• and
• a region extraction step of separately extracting, by a region extraction unit, a candidate region estimated to contain the linear mucosal structure and a candidate region estimated to contain the massive mucosal structure based on a calculation result of the geometric evaluation value.