Apparatus and method for identification of tissue regions in digital mammographic images

Claim 3

The method of claim 1 wherein the amount of image reduction in said scaling step is matched to the spatial resolution of said provided digital image and the filter size used in said texture extracting step.

Claim 4

The method of claim 1 wherein said texture extraction system of said texture extracting step includes convolving said scaled image with a spatial filter.

Claim 5

The method of claim 3 wherein said spatial filter uses a 5×5 pixel kernel derived from center-weighted vectors defined by the Laws texture measure, said vectors being L5= 1 4 6 4 1! E5= -1 -2 0 2 1! S5= -1 0 2 0 -1! W5= -1 2 0 -2 1! R5= 1 -4 6 -4 1!

Claim 6

The metod of claim 4 wherein said spatial filter uses four 5×5 kernels derived from multiplying the following combination of vectors E5E5, S5S5, 35S5, S5E5, such that each texure image corresponds to one of said four kernels.

Claim 7

The method of claim 1 wherein said clustering system of said clustering step uses a K-means clustering algorithm to convert each said texture image into vectors of equal size which is determined by the number of texture images.

Claim 8

The method of claim 1 wherein said labeling system of said labeling step connect similar regions of said image into contiguous regions through the splitting and merging algorithm.

Claim 9

The method of claim 1 wherein said thresholding step sets the background pixels of said digital image to zero and the remaining region pixels to a high code value.

Claim 10

A computer storage product comprising: a computer readable storage medium having a computer program stored thereon for performing a method of identifying tissue regions in a digital image representing a body part comprising the steps of: providing a digital image representing a body part having tissue regions; scaling said provided digital image by subsampling or interpolating, to produce a scaled digital image; texture extracting said scaled digital image by texture extraction system to produce a plurality of texture images having a different measure of image texture of said scaled digital image; clustering said plurality of texture images by operation of a clustering system to produce a clustered image; labeling said clustered image by operation of a labeling system based on both pixel intensity values and geometric adjacency of the pixels to produce a labeled image; thresholding the labeled image by operation of thresholding system to produce a threshold image; filtering said threshold image by operation of a largest component filter system to produce a largest component image which is the largest geometrically connected region in said threshold image; edge detecting said largest component image by operation of an edge detection system to produce an edge image containing only edge pixels at high code value; operating on said edge image by a convex hull calculation system to produce a convex hull image which outlines the skin line of the tissue region of said body part of said provided digital image.

Claim 11

A computer storage product comprising: a computer readable storage medium having a computer program stored thereon for performing a method of identifying tissue regions in a digital image representing a body part comprising the steps of: providing a digital image representing a body part having tissue regions; scaling said provided digital image by subsampling or interpolating, to produce a scaled digital image; texture extracting said scaled digital image by texture extraction system to produce a plurality of texture images having a different measure of image texture of said scaled digital image; clustering said plurality of texture images by operation of a clustering system to produce a clustered image; labeling said clustered image by operation of a labeling system based on both pixel intensity values and geometric adjacency of the pixels to produce a labeled image; thresholding the labeled image by operation of thresholding system to produce a threshold image; filtering said threshold image by operation of a largest component filter system to produce a largest component image which is the largest geometrically connected region in said threshold image; edge detecting said largest component image by operation of an edge detection system to produce an edge image containing only edge pixels at high code value; operating on said edge image by a convex hull calculation system to produce a convex hull image which outlines the skin line of the tissue region of said body part of said provided digital image.