System and method for salient region feature based 3D multi modality registration of medical images
AssigneeSiemens Medical Solutions USA, Inc.
InventorsChenyang Xu, Dieter Hahn, Yiyong Sun, Frank Sauer
Life of this patent
Estimating the term…
Abstract
A method for aligning a pair of images includes providing a pair of images, identifying salient feature regions in both a first image and a second image, wherein each region is associated with a spatial scale, representing feature regions by a center point of each region, registering the feature points of one image with the feature points of the other image based on local intensities, ordering said feature pairs by a similarity measure, and optimizing a joint correspondence set of feature pairs by refining the center points to sub-pixel accuracy.
Claims25
A method of aligning a pair of images comprising at least one computer implementing the steps of:
- providing a pair of images with a first image and a second image, wherein said images comprise a plurality of intensities corresponding to a domain of pixels in a 3-dimensional space,
- identifying salient feature regions in both the first image and the second image, wherein each region is associated with a spatial scale;
- representing feature regions by a center point of each region;
- registering the feature points of one image with the feature points of the other image based on local intensities;
- ordering said feature pairs by a similarity measure;
- and
- optimizing a joint correspondence set of feature pairs by refining the center points to sub-pixel accuracy wherein representing salient feature region center points for one image in a KD-tree, querying said KD-tree for each feature to find a set of nearest neighbor features, and removing from said tree those nearest neighbor features with a lower saliency value and that have a center point within the scale of said each feature, wherein a substantially uniform distribution of salient feature regions in said image is achieved.
A method of aligning a pair of images comprising at least one computer implementing the steps of:
- providing a pair of images with a first image and a second image, wherein said images comprise a plurality of intensities corresponding to a domain of pixels in a 3-dimensional space,
- identifying salient feature regions in both the first image and the second image, wherein each region is associated with a spatial scale;
- estimating an initial registration using an iterative closest point transformation between said first image and said second image;
- transforming all features of said second image into a coordinate space of said first image;
- storing said transformed features in a kD-tree, and querying said tree for each feature in said first image to select those nearest neighbor features in said second image based on a predefined selection criteria;
- testing said selected feature pairs of features in said first image and said second image for translation invariance, rotation invariance, and a global image similarity measure;
- and
- ordering said selected feature pairs by their global image similarity measure values.
A program storage device readable by a computer, tangibly embodying a program of instructions executable by the computer to perform the method steps for aligning a pair of images, said method comprising the steps of:
- providing a pair of images with a first image and a second image, wherein said images comprise a plurality of intensities corresponding to a domain of pixels in a 3-dimensional space,
- identifying salient feature regions in both the first image and the second image, wherein each region is associated with a spatial scale;
- representing feature regions by a center point of each region;
- registering the feature points of one image with the feature points of the other image based on local intensities;
- ordering said feature pairs by a similarity measure;
- and
- optimizing a joint correspondence set of feature pairs by refining the center points to sub-pixel accuracy wherein representing salient feature region center points for one image in a KD-tree, querying said KD-tree for each feature to find a set of nearest neighbor features, and removing from said tree those nearest neighbor features with a lower saliency value and that have a center point within the scale of said each feature, wherein a substantially uniform distribution of salient feature regions in said image is achieved.
Dependencies are inferred from the claim text.