System and method for boundary detection in tomographic images by geometry-constrained edge detection of projected data
US6026142A · kind A · utility
Assignee
Inventors
Key dates
| Filing date | Apr 2, 1998 |
| Grant date | Feb 15, 2000 |
| Priority date | — |
| Expiry date | Apr 2, 2018 |
Classification
- Technology area (CPC Y)Emerging Cross-Sectional Technologies
- CPC primaryY10S378/901
- WIPO fieldComputer technology
- WIPO sectorElectrical engineering
Abstract
A system and method extracts object boundaries in computed tomography images that have been corrupted by noise. The system and method find object boundaries in tomography images by finding edges in sinogram data, thereby providing a segmentation method that is insensitive to the image noise introduced by the process of image reconstruction from projections. The boundary-detection system uses projection data, and optionally tomographic image data, to detect the boundaries of objects in tomographic images. The system is of particular value for boundary-detection in images corrupted by reconstruction artifacts, such as CT scans of hospital patients who have implanted metal prostheses. The system is applicable to recovering the boundaries of convex objects, and the convex segments of the boundaries of concave objects (though not the complete boundaries of the concave objects themselves). The system is clinically viable in the sense that this method can be embodied in a practical, real-world system that can be used routinely in hospitals and medical clinics, and relies only on data that are available from standard medical CT scanners, that is the CT images and scout images (and not the …
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.