Clavicle suppression in radiographic images
View Patent ↗A method for clavicle suppression in a chest x-ray image. The method identifies the lung fields in the x-ray image and detects at least one portion of a clavicle ridge that lies outside the lung fields. Edges of a clavicle on each side of the detected clavicle ridge are detected, edge detection extended for the clavicle edges into the lung fields, and the clavicle defined within the x-ray image according to the edge detection. The clavicle is suppressed within the x-ray image to generate a clavicle-suppressed x-ray image and the clavicle-suppressed x-ray image is displayed, stored, or transmitted.
1. A method for processing a chest x-ray image, executed at least in part by a computer, comprising:
identifying lung fields in the x-ray image;
detecting at least one portion of a clavicle that lies outside the lung fields by: (i) generating one or more gradient feature images, (ii) applying a gradient analysis to the one or more gradient feature images to detect a set of candidate portions, wherein the gradient analysis includes an assessment of at least one of the following: gradient strength, gradient length, gradient angle, gradient intensity, and symmetry factors, and (iii) applying a false-positive analysis to the set of candidate portions;
detecting a clavicle ridge based on the detected clavicle portion;
detecting clavicle edges of the detected clavicle portion;
guided by the clavicle ridge, extending edge detection for the clavicle edges into the lung fields to detect clavicle line segments within the lung fields;
defining the clavicle within the x-ray image according to the edge detection;
suppressing the clavicle within the x-ray image to generate a clavicle-suppressed x-ray image; and
displaying, storing, or transmitting the clavicle-suppressed x-ray image.
2. The method of claim 1 wherein detecting the at least one portion of the clavicle comprises applying a derivative analysis filter.
3. The method of claim 2 wherein applying a derivative analysis filter includes using a Laplacian filter.
4. The method of claim 1 wherein detecting the at least one portion of the clavicle comprises applying results from trained system processing.
5. The method of claim 1 further comprising refining the edges.
6. The method of claim 1 further comprising generating and applying an artifact mask.
7. The method of claim 1 wherein detecting the at least one portion of the clavicle further comprises identifying and removing false positives.
8. The method of claim 1 wherein detecting edges of the clavicle further comprises checking symmetry.
9. A method for processing a chest x-ray image, executed at least in part by a computer, comprising:
identifying a region of interest that includes lung fields in the chest x-ray image;
detecting at least one portion of a clavicle that lies within the region of interest and outside the lung fields by: (i) generating one or more gradient feature images, (ii) applying a gradient analysis to the one or more gradient feature images to detect a set of candidate portions, wherein the gradient analysis includes an assessment of at least one of the following: gradient strength, gradient length, gradient angle, gradient intensity, and symmetry factors, and (iii) applying a false-positive analysis to the set of candidate portions;
detecting a clavicle ridge based on the detected clavicle portion;
forming a clavicle mask that includes the at least one portion of the detected clavicle;
detecting and refining edges of the detected at least one portion of the clavicle, wherein the edges lie within the clavicle mask;
guided by the detected clavicle ridge, extending edge detection for the detected clavicle edges into the lung fields and refining the clavicle mask to include edges that lie within the lung fields;
suppressing the clavicle within the clavicle mask to generate a clavicle-suppressed x-ray image; and
displaying, storing, or transmitting the clavicle-suppressed x-ray image.
10. The method of claim 9 wherein detecting the at least one portion of the clavicle comprises applying a trained system to classify pixels in the region of interest as clavicle or non-clavicle.
11. The method of claim 9 further comprising generating and applying an artifact mask over an area of the chest x-ray image that includes tubing or wire.
12. The method of claim 9 wherein refining the clavicle mask further comprises analyzing at least one of gradient strength, gradient length, or gradient angle for the edges that lie within the lung fields.
13. The method of claim 9 wherein detecting edges of the at least one portion of the clavicle further comprises identifying and removing false positives.
14. A method for processing a chest x-ray image, executed at least in part by a computer, comprising:
segmenting lung fields in the x-ray image;
generating and refining a clavicle search mask according to the segmented lung fields, wherein the clavicle search mask includes a portion of the image lying outside the segmented lung fields;
applying gradient analysis to portions of the image corresponding to the generated clavicle search mask and identifying a clavicle ridge and one or more clavicle edges that lie within the clavicle search mask and outside the lung fields;
guided by the clavicle ridge, extending the one or more identified clavicle edges into the lung fields and identifying one or both clavicles;
suppressing the one or both clavicles within the chest x-ray image to generate a clavicle-suppressed x-ray image; and
displaying, storing, or transmitting the clavicle-suppressed x-ray image.
15. The method of claim 14 wherein the step of applying gradient analysis includes using a Laplacian filter.