System and method for PDE-based multiphase segmentation
AssigneeSiemens Medical Solutions USA, Inc.
InventorsSheshadri Thiruvenkadam, Christophe Chefd'hotel
Life of this patent
Estimating the term…
Abstract
A method for segmenting a digitized medical image including providing a digitized image comprising a plurality of intensities corresponding to a domain of points on a n-dimensional grid, modeling the segmentation phases of said image with 2k weighting functions defined as a product of k unknown auxiliary functions, minimizing an energy functional of said weighting functions and said auxiliary functions to derive k gradient flow equations for said auxiliary functions, numerically integrating said gradient flow equations wherein solutions for the k unknown auxiliary functions are obtained, and partitioning said image into segments based on a membership of each image point in each of said weighting functions.
Claims22
A method of segmenting a digitized medical image comprising the steps of:
- providing a digitized image comprising a plurality of intensities corresponding to a domain of points on a n-dimensional grid;
- modeling the segmentation phases of said image with 2k weighting functions defined as a product of k unknown auxiliary functions;
- minimizing an energy functional of said weighting functions and said auxiliary functions to derive k gradient flow equations for said auxiliary functions;
- numerically integrating said gradient flow equations wherein solutions for the k unknown auxiliary functions are obtained;
- and
- partitioning said image into segments based on a membership of each image point in each of said weighting functions.
A method of segmenting a digitized medical image comprising the steps of:
- providing a digitized image comprising a plurality of intensities corresponding to a domain of points on a n-dimensional grid;
- modeling the segmentation phases of said image with 2k weighting functions defined as a product of k unknown auxiliary functions;
- and
- minimizing an energy functional of said weighting functions and said auxiliary functions to derive k equations for said auxiliary functions, wherein said image is partitioned into 2k segments Ωi based on a membership of each image point in each of said weighting functions according to Ωi={x|Πj=1,j≠i2k(Ai−Aj)>0}, wherein x is a point in said image domain, and Ai and Aj are weighting functions.
A program storage device readable by a computer, tangibly embodying a program of instructions executable by the computer to perform the method steps for segmenting a digitized medical image comprising the steps of:
- providing a digitized image comprising a plurality of intensities corresponding to a domain of points on a n-dimensional grid;
- modeling the segmentation phases of said image with 2k weighting functions defined as a product of k unknown auxiliary functions;
- minimizing an energy functional of said weighting functions and said auxiliary functions to derive k gradient flow equations for said auxiliary functions;
- numerically integrating said gradient flow equations wherein solutions for the k unknown auxiliary functions are obtained;
- and
- partitioning said image into segments based on a membership of each image point in each of said weighting functions.
Dependencies are inferred from the claim text.