Method and Apparatus for Segmenting Small Structures in Images

What is needed is a segmentation method and apparatus without statistical models, local statistics, or thresholds to be selected manually, and with significantly lower computational complexity compared to the multi-tolerance and active contours methods, for enhanced speed.

In particular, what is needed is a method and apparatus to segment pixels in an image, such as a mammogram, containing a plurality of extra dark or extra bright objects just a few pixels in extent, that gives edges similar to those selected by an expert, but does so with fewer computations and with fewer manually adjustable parameters than conventional segmentation methods and equipment.

Therefore it is an object of the present invention to provide segmentation for small features in multidimensional data which defines small feature edges that correspond closely to those selected by an analyst but does so with less complexity than the above known methods.

It is another object of the present invention to provide a data processing apparatus that more rapidly provides small feature edges that correspond closely to those selected by an analyst.

It is another object of the present invention to provide computer program products that more rapidly provide small feature edges that correspond closely to those selected by an analyst.

It is another object of the invention to identify microcalcifications in a mammogram.

These and other objects and advantages of the present invention are provided by a method for segmenting a small feature in a multidimensional digital array of intensity values in a data processor. Each small feature includes a local intensity extreme, such as an intensity maximum. An edge metric is computed along each ray of a plurality of multidimensional rays originating at the local intensity extreme. A multidimensional edge point is identified corresponding to a maximum edge metric on each ray. Every point on each ray from the local extreme to the ray edge point is labeled as part of the small feature. The labeling is then spread to an unlabeled point following a hill climbing procedure requiring that the unlabeled point be adjacent to a labeled point, have a similar or more extreme intensity than the labeled point, and be closer than the labeled point to the local extreme.

In another embodiment, the multidimensional array is a digital image, and each point is a pixel. In another embodiment, the digital image is a digitized mammogram and the small feature is a microcalcification candidate. In the latter embodiment, microcalcification candidates are satisfactory segmented in fewer operations than with conventional segmentation methods.

In another aspect of the invention, a data processing apparatus segments a small feature in a multidimensional digital array of intensity values. The apparatus includes an input for inputting a plurality of intensity values arranged along regular increments in each of a plurality of dimensions and a memory medium for storing the plurality of intensity values as a multidimensional digital array. The apparatus includes a processor configured to detect a local intensity extreme in the multidimensional digital array, to identify points along a plurality of rays originating at the local intensity extreme, and to identify one ray edge point on each ray. The ray edge point is associated with a maximum edge metric along the ray. The processor is also configured to label the points in the array that are part of the small features. Each point on each ray from the local intensity extreme to the edge point is labeled, as is an unlabeled point adjacent to a labeled point if the unlabeled point has a more extreme intensity than the labeled point and the unlabeled point is closer than the labeled point to the local extreme. Labeling continues until no more unlabeled points can be labeled. The apparatus also includes an output for providing the labeled points for subsequent processing.

In another aspect of the invention, a computer program product is provided for segmenting a small feature in a multidimensional array of intensities using a computer. The computer program product includes computer controlling instructions for configuring a computer to compute an edge metric along each ray of a plurality of multidimensional rays originating at a local intensity extreme. The instructions also identify a ray edge multidimensional point corresponding to a maximum edge metric on each ray. The program also labels every point on each ray from the local extreme to the ray edge point, and then labels an unlabeled point if the unlabeled point is adjacent to a labeled point and the unlabeled point has a more extreme intensity than the labeled point, and the unlabeled point is closer than the labeled point to the local extreme. In one embodiment, the instructions are stored in a computer readable memory device. In another embodiment, the instructions are transmitted as electronic signals on a communications line.

The foregoing and other features, aspects and advantages of the present invention will become more apparent from the following detailed description of the present invention when taken in conjunction with the accompanying drawings.

Patents:
US 7,106,893

Type of Offer: Licensing



Next Patent »
« More Software Patents

Share on      


CrowdSell Your Patent