Reducing false positives for automatic computerized detection of objects
View Patent ↗A computer-implemented method for identifying an object of interest includes providing input data including an image and a candidate for the object of interest in the image, extracting a boundary of the candidate, and extracting a segment of a region of interest containing the candidate. The method further includes determining a plurality of features of an extracted segment of the region of interest containing the candidate, and outputting the object of interest, wherein the object of interest is characterized by the plurality of features, wherein the object of interest and the plurality of features are stored as computer-readable code.
1. A computer-implemented method for automatically identifying a pulmonary embolism comprising:
acquiring input data including a digital or digitized medical image, the image comprising:
a vessel; and
a candidate in the vessel;
automatically extracting a boundary of the candidate;
automatically extracting a segment of the vessel containing the candidate, wherein extracting the segment of the vessel comprises seeding the segment extraction using the boundary, and growing the boundary;
automatically determining a plurality of features of the extracted segment of the vessel containing the candidate;
automatically identifying the candidate as a pulmonary embolism based upon the plurality of features of the extracted segment of the vessel including comparing the plurality of features to a known set of pulmonary embolism features and verifying the pulmonary embolism based on the comparison; and
automatically outputting data identifying the candidate as a pulmonary embolism and the plurality of features, wherein the data and the plurality of features are stored as computer-readable code.
2. The computer-implemented method of claim 1 , wherein automatically extracting the segment of the vessel comprises providing the segment of the vessel having a fixed size.
3. The computer-implemented method of claim 1 , wherein acquiring the input data comprises:
acquiring input data including a digital or digitized medical image, the image comprising the candidate as a plurality of voxels;
providing a value for stopping a region growing as a maximum distance between a voxel and the boundary of the candidate; and
providing a threshold for an intensity of the voxel.
4. The computer-implemented method of claim 1 , wherein automatically outputting further comprises classifying the candidate as a false positive or a true positive based on the plurality of features.
5. The computer-implemented method of claim 1 , wherein the plurality of features includes a texture feature.
6. The computer-implemented method of claim 1 , wherein the plurality of features includes a histogram feature.
7. The computer-implemented method of claim 1 , wherein the plurality of features includes an intensity feature.
8. The computer-implemented method of claim 1 , wherein the plurality of features includes a difference feature.
9. The computer-implemented method of claim 1 , wherein the plurality of features includes a curvature feature.
10. A computer-implemented method for automatically identifying a pulmonary embolism in a digital or digitized medical image, comprising:
acquiring a digital or digitized medical image, the medical image including a vessel;
automatically identifying a candidate within the image;
automatically identifying a boundary of the candidate;
automatically segmenting the vessel, wherein segmenting the vessel comprises seeding the segmenting using the candidate boundary and growing the boundary;
automatically analyzing and determining a plurality of features of the segmented vessel; and
automatically identifying the candidate as a pulmonary embolism based upon the plurality of features of the segmented vessel including comparing the plurality of features to a known set of pulmonary embolism features and verifying the pulmonary embolism based on the comparison.
11. The computer-implemented method of claim 10 , wherein automatically segmenting the vessel comprises providing a segment of the vessel having a fixed size.
12. The computer-implemented method of claim 10 , wherein acquiring the input data comprises:
acquiring input data including a digital or digitized medical image, the image comprising the candidate as a plurality of voxels;
providing a value for stopping a region growing as a maximum distance between a voxel and the boundary of the candidate; and
providing a threshold for an intensity of the voxel.
13. The computer-implemented method of claim 10 , wherein outputting the pulmonary embolism further comprises classifying the candidate as a false positive or a true positive based on the plurality of features.
14. The computer-implemented method of claim 10 , wherein the plurality of features includes a texture feature.
15. The computer-implemented method of claim 10 , wherein the plurality of features includes a histogram feature.
16. The computer-implemented method of claim 10 , wherein the plurality of features includes an intensity feature.
17. The computer-implemented method of claim 10 , wherein the plurality of features includes a difference feature.
18. The computer-implemented method of claim 10 , wherein the plurality of features includes a curvature feature.