Positioning method of multiple organs in three-dimensional image

Abstract

The invention provides a positioning method of multiple organs in a three-dimensional image. The positioning method comprises constructing a coexistence generalized Hough transform lookup table of the offset of a boundary point of any one organ relative to the positions of the center of mass of all the organs based on an organ template training method; inputting a three-dimensional image, and determining positions and gradients of the boundary points of all the organs in the three-dimensional image; and based on the positions and gradients of the boundary points and the lookup table, determining the position of the center of mass of each organ in the three-dimensional image, that is determining the position of each organ in the three-dimensional image. According to the invention, by use of the form information on an organ in need of position and relative position relation information with other organs, the position of the organ in the image can be quickly, accurately and stably positioned.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a multi-organ positioning method in a three-dimensional image. Background technique [0002] Fast, accurate and stable localization of human organs in medical images is usually an important step in computer-aided diagnosis. The automatic segmentation of organs in medical images, the extraction and quantitative analysis of organ structures (including lesions) all depend on accurately locating the location of each organ in complex medical images. [0003] Organ localization methods in medical images can be roughly classified into two categories. The first type of method usually does not require any prior organ model, but searches a large number of medical images as training samples, extracts the features of each organ in the image through machine learning, and determines the organ of interest based on these features Position in an unknown image [S.Seifert, A.Barbu, S.K.Zhou, D.Liu...

AI Technical Summary

Problems solved by technology

However, it is worth noting that the generalized Hough transform currently used in organ localization only uses the morphological model of the organ to be localized, and discards the important context information formed by the organ and other nearby major organs.

Therefore, when organs with similar local shapes appear in the image, the generalized Hough transform is prone to positioning errors

Images