Back to results
US 2016/0260212 A1Published application
Vessel segmentation method
AssigneeAgfa Healthcare NV
InventorsAsma OUJI, Yoni DE WITTE
Life of this patent
Estimating the term…
Abstract
Vascular structures are segmented by a locally adaptive method based on iterative region growing and morphological operations.
Claims6
1-5. (canceled)
A method of segmenting a structure in a volume represented by a 3D digital representation, the method comprising the steps of:
- applying a region growing algorithm to the volume starting from a seed point to generate a final mask;
- evaluating the final mask to identify one or more regions including unconnected structures;
- when a region of the one or more regions including unconnected structures is found, identifying a new seed point in the region;
- applying the region growing algorithm to the new seed point resulting in a seed mask;
- and
- when the seed mask intersects with the final mask, creating a mask as a combination of the seed mask and the final mask, otherwise, applying a locally adaptive region growing iteration including iteratively performing the region growing algorithm to a next seed point to generate a new mask until the new mask intersects the final mask;
- wherein
- the locally adaptive region growing iteration includes:
- determining a first gray value distance between the next seed point and a low threshold value used to generate the seed mask and a second gray value distance between the next seed point and a high threshold value used to generate the seed mask;
- determining a minimum distance between the first and second gray value distances;
- when the minimum distance is lower than a preset threshold, enlarging a range between the low and the high threshold values so as to encompass the next seed point by replacing the low threshold value by a minimum of the low threshold value and the gray value of the next seed point and by replacing the high threshold value by a maximum of the high threshold and the gray value of the next seed point, otherwise replacing the next seed point by the new seed point;
- and
- applying the region growing algorithm to the volume starting from the new seed point to generate the new mask.
10 more elements
Dependencies are inferred from the claim text.