Tomosynthesis reconstruction with rib suppression
View Patent ↗A method for rib suppression in a volume chest x-ray image, executed at least in part by a computer captures a first set of unsegmented projection images, each at a corresponding angle, and forms a second set of segmented projection images by detecting rib features in a first projection image to form a first segmented projection image and generating a base model according to the detected rib features for the first projection image. Each of one or more additional projection images from the first set is processed to add members to the second set by a repeated sequence of generating a predictive model; detecting rib features using the predictive model; adjusting the base model according to detected rib features; and correcting rib detection in one or more members of the second set. The volume chest x-ray image is reconstructed according to the segmented projection images and is displayed.
1. A method for rib suppression in a volume chest x-ray image, the method executed at least in part by a computer system, comprising:
forming a first set of unsegmented projection images by capturing a plurality of projection images, each projection image having a corresponding capture angle;
forming a second set of segmented projection images from the first set of unsegmented projection images by:
(a) detecting rib features in a first unsegmented projection image to form a first segmented projection image in the second set and generating a base model according to the detected rib features for the first unsegmented projection image;
(b) processing each of one or more additional unsegmented projection images from the first set to add members to the second set by a repeated sequence of:
(1) generating a predictive model from the base model;
(2) detecting rib features in the unsegmented projection image using the generated predictive model to store a segmented image in the second set of images;
(3) adjusting the base model according to the detected rib features; and
(4) correcting rib detection in one or more members of the second set of images;
conditioning one or more of the unsegmented projection images of the first set of images to suppress rib content based on information from segmentation of the second set of images;
reconstructing the volume chest x-ray image according to two or more of the conditioned projection images; and
displaying, storing, or transmitting the reconstructed volume chest x-ray image.
2. The method of claim 1 wherein detecting rib features for unsegmented projection images comprises forming a conditioned projection image by:
identifying a region of interest in the projection image;
detecting and labeling one or more ribs in the region of interest;
detecting rib edges of the one or more detected ribs; and
generating cross rib profiles along the detected ribs.
3. The method of claim 1 further comprising generating a correspondence between one or more adjacent images using incremental shifting.
4. The method of claim 1 wherein detecting rib features comprises identifying the lung region.
5. The method of claim 1 wherein detecting rib features comprises extracting a medial axis from at least one of the one or more imaged ribs.
6. The method of claim 1 wherein detecting rib features comprises adjusting segments of one or more of detected rib edges to improve edge fitting.
7. The method of claim 1 further comprising scaling one or more unsegmented projection images to a reduced resolution.
8. The method of claim 1 wherein detecting rib features comprises applying a growing or extending algorithm for rib edges.
9. The method of claim 1 wherein the second set of images has fewer members than the first set of images.
10. A method for rib suppression in a volume chest x-ray image, the method executed at least in part by a computer system and comprising:
forming a first set of unsegmented projection images by capturing a plurality of projection images, each at a corresponding capture angle;
forming a second set of segmented projection images from the first set of unsegmented projection images by a repeated sequence of:
(a) designating a first projection image from among members of the first set of images that are not yet segmented;
(b) detecting rib features in the first projection image and in at least one adjacent projection image that was captured before or following the first projection image;
(c) refining the rib feature detection for the first projection image and the at least one adjacent projection image according to the rib feature detection for the first projection image and the at least one adjacent projection image to generate a first segmented projection image; and
(d) assigning the first segmented projection image to the second set of images;
suppressing rib feature content from one or more of the unsegmented projection images of the first set of images based on the segmentation information from the second set of images;
reconstructing the volume chest x-ray image according to the plurality of rib-suppressed projection images; and
displaying, storing, or transmitting the reconstructed, volume chest x-ray image.
11. The method of claim 10 wherein detecting rib features for unsegmented projection images comprises forming a conditioned projection image by:
identifying a region of interest in the projection image;
detecting and labeling one or more ribs in the region of interest;
detecting rib edges of the one or more detected ribs; and
generating cross rib profiles along the detected ribs.
12. The method of claim 10 further comprising conditioning one or more of the projection images in the second set to suppress rib content.
13. The method of claim 10 wherein detecting rib features comprises identifying the lung region.
14. The method of claim 10 wherein detecting rib features comprises extracting a medial axis from at least one of the one or more imaged ribs.
15. The method of claim 10 wherein detecting rib features comprises adjusting segments of one or more of detected rib edges to improve edge fitting.
16. The method of claim 10 further comprising scaling one or more unsegmented projection images to a reduced resolution.
17. The method of claim 10 wherein detecting rib features comprises applying a growing or extending algorithm for rib edges.
18. A method for rib suppression in a volume chest x-ray image, the method executed at least in part by a computer system and comprising:
forming a first set of unsegmented projection images by capturing a plurality of projection images, each at a corresponding capture angle;
forming a second set of segmented projection images from the first set of unsegmented projection images by:
(a) detecting ribs in a subset of the unsegmented projection images;
(b) segmenting rib content within the subset of images;
(c) registering the subset of images to each other;
(d) generating a base model according to the registered subset of images;
(e) processing each of one or more additional unsegmented projection images from the first set to add members to the second set by a repeated sequence of:
(1) generating a predictive model from the base model;
(2) detecting rib features in the unsegmented projection image using the generated predictive model to store a segmented image in the second set;
(3) adjusting the base model according to the detected rib features; and
(4) correcting rib detection in one or more members of the second set;
conditioning one or more of the unsegmented projection images of the first set of images to suppress rib content based on information from segmentation of the second set of images;
reconstructing the volume chest x-ray image according to two or more of the conditioned projection images; and
displaying, storing, or transmitting the reconstructed volume chest x-ray image.
19. The method of claim 18 further comprising storing at least the first set of unsegmented projection images.