US 9,519,966 B2 · Method, system and computer program product for breast density classification using parts-based local features

Abstract

An automated content-based image retrieval method, a system and a computer program product for the classification of breast density from mammographic imagery. Raw digital mammogram images taken of patients are initially pre-processed to remove noise and enhance contrast, then subjected to pectoral muscle segmentation to produce region of interest (ROI) images. The ROI images are then decomposed using non-negative matrix factorization (NMF), where a non-negative sparsity constraint and reconstruction quality measures are imposed on the extracted and retained first few NMF factors. Based on the retained NMF factors, kernel matrix-based support vector machines classify the mammogram images binomially or multinomially to breast density categories. Methods of assessing and comparing the NMF-based breast classification method to principal component analysis or PCA-based methods are also described, and the NMF-based method is found to achieve higher classification accuracy and better handling of invariance in the digital mammogram images because of its parts-based factorization.

Claims14

Dependent claims

A computer-implemented method for classifying breast density from mammographic imagery, the method comprising:

pre-processing one or more digital mammogram images of a patient with a processing circuitry to remove noise and enhance contrast;
segmenting pectoral muscle from the one or more pre-processed digital mammogram images to produce one or more extracted region of interest images and saving the one or more extracted region of interest images in a computer memory;
grouping the one or more extracted region of interest images to produce one large mammogram image;
decomposing a m-by-n mammogram image matrix A associated with the large mammogram image with the processing circuitry by non-negative matrix factorization to extract and retain a plurality of non-negative matrix factors;
and
classifying the plurality of non-negative matrix factors with a kernel matrix-based support vector machine according to breast density to two or more breast density categories;
wherein the non-negative matrix factorization is according to one of the following equations:
Am×n≈Wm×rPr×x (Eq. 2);
or
Am×n=Wm×rPr×n+Em×n (Eq. 3);
where columns of W are NMF base images, rows of P are encoding coefficients of the NMF base images, E is factorization error, r is rank and such that (m+n)r<m×n.

8 more elements

A non-transitory computer-readable storage medium having computer-readable instructions thereon which, when executed by a computer, causes the computer to perform a computer-implemented method for classifying breast density from mammographic imagery, the method comprising:

pre-processing one or more digital mammogram images of a patient with a processing circuitry to remove noise and enhance contrast;
segmenting pectoral muscle from the one or more pre-processed digital mammogram images to produce one or more extracted region of interest images and saving the one or more extracted region of interest images in a computer memory;
grouping the one or more extracted region of interest images to produce one large mammogram image;
decomposing a m-by-n mammogram image matrix A associated with the large mammogram image with the processing circuitry by non-negative matrix factorization to extract and retain a plurality of non-negative matrix factors;
and
classifying the plurality of non-negative matrix factors with a kernel matrix-based support vector machine according to breast density to two or more breast density classes;
wherein the non-negative matrix factorization is according to one of the following equations:
Am×n≈Wm×rPr×x (Eq. 2);
or
Am×n=Wm×rPr×n+Em×n (Eq. 3);
where columns of W are NMF base images, rows of P are encoding coefficients of the NMF base images, E is factorization error, r is rank and such that (m+n)r<m×n.

8 more elements

Dependencies are inferred from the claim text.

Method, system and computer program product for breast density classification using parts-based local features

Life of this patent

Method, system and computer program product for breast density classification using parts-based local features

Life of this patent

Abstract

Claims14