Mass spicules detection and tracing from digital mammograms
AssigneeThree Palm Software
InventorsHeidi Daoxian Zhang, Patrick Bernard Heffernan
Life of this patent
Estimating the term…
Abstract
The present invention provides an algorithm to detect and trace the spicules of a mass density in digital mammograms using an adaptive threshold edge algorithm and a flood-fill segmentation algorithm. Elongation criteria are used to remove false edges that do not radiate from a central mass margin. The algorithm works on a central mass border and spicules feature map that contains a subset of the pixels from the source image, so processing time is fast enough for use in a mammography CAD server and for real-time computation within a digital mammography workstation.
Claims13
A method of producing spicule traces from digital mammogram images, comprising:
- selecting mass density candidates to obtain regions of interest (ROIs);
- pre-processing each mass density candidate to obtain a central mass segmentation and a corresponding spicules features map, wherein pre-processing each mass density candidate comprises:
- extracting a region of interest at each mass density segmentation candidate location;
- resampling the extracted region to a fixed dimension;
- enhancing the extracted region to produce an enhanced image;
- performing flood-fill segmentation on the enhanced image to search for a central mass segmentation;
- and
- applying an adaptive threshold to the enhanced image and an edge image to produce the spicules features map;
- and
- tracing the obtained central mass segmentations and associated spicules to produce the spicule traces.
A method of producing spicule traces from digital mammogram images, comprising:
- selecting mass density candidates to obtain regions of interest (ROIs);
- pre-processing each mass density candidate to obtain a central mass segmentation and a corresponding spicules features map, wherein pre-processing each mass density candidate comprises:
- extracting a region of interest at each mass density segmentation candidate location, the extracted region of interest being four times the estimated mass size at each mass density segmentation candidate location;
- resampling the extracted region to a fixed dimension, wherein resampling the extracted region comprises upsampling or downsampling the region to a dimension of 256 by 256 pixels;
- enhancing the extracted region to produce an enhanced image, wherein enhancing the extracted region comprises:
- bandpass filtering the extracted region using a Gaussian blur difference filter to produce a bandpass image;
- increasing a weighting of the bandpass image by a predetermined amount;
- adding the weighted bandpass image and the edge image, to produce a combined image;
- and
- modulating the combined image with a modulator and Gaussian weights to obtain the enhanced image;
- performing flood-fill segmentation on the enhanced image to search for a central mass segmentation, wherein performing a flood-fill segmentation comprises performing a flood-fill in a range between a center of the enhanced image and a maximum pixel value within the enhanced image;
- and
- applying an adaptive threshold to the enhanced image and an edge image to produce the spicules features map;
- and
- tracing the obtained central mass segmentations and associated spicules to produce the spicule traces.
A method of producing spicule traces from digital mammogram images, comprising:
- selecting mass density candidates to obtain regions of interest (ROIs);
- pre-processing each mass density candidate to obtain a central mass segmentation and a corresponding spicules features map;
- and
- tracing the obtained central mass segmentations and associated spicules to produce the spicule traces, wherein tracing the obtained central mass segmentations and associated spicules comprises:
- dilating the obtained central mass segmentations by one pixel;
- subtracting the obtained central mass segmentations from the dilated mass segmentations to obtain a central mass border trace;
- masking the central mass segmentations with the spicules feature map;
- adding the masked result to the central mass border trace to obtain a hollow spicules feature map;
- selecting a pixel on the central mass border trace as a seed;
- applying a flood-fill over the hollow spicules feature map to obtain the spicules feature segmentation;
- subtracting the central mass border trace from the spicules feature segmentation to obtain individual spicules segmentation masks;
- and
- using elongation criteria to remove false spicules that doe not radiate from a centroid of the central mass segmentation.
Dependencies are inferred from the claim text.