Computer aided diagnosis using dual energy subtraction images
View Patent ↗Computer-aided diagnosis techniques may be combined with dual-energy radiography techniques to provide enhanced computer-aided diagnosis of, for example, lung nodules.
1. A method for performing detection based on radiographic images, the method comprising:
processing at least two radiographic images obtained using different energy levels to obtain derived images at least a soft-tissue equivalent image, a bone equivalent image, and an image including both soft tissue and bone;
processing at least one of said derived images to detect at least one feature within said at least one derived image, wherein said processing at least one of said derived images comprises at least two processes selected from the group consisting of multi-resolution enhancement, object feature extraction, linear neural network classification, and fuzzy neural network classification, and wherein said processing at least one of said derived images further comprises performing decision fusion based on results obtained from said at least two processes; and
processing said at least one feature using decision fusion to obtain one or more final detection results.
2. The method according to claim 1 , wherein said processing at least two radiographic images comprises:
performing a separation algorithm to identify image portions comprising mainly soft tissue and image portions comprising mainly bone.
3. The method according to claim 1 , wherein said processing at least one of said derived images further comprises:
performing image segmentation.
4. The method according to claim 1 , wherein said processing at least one feature comprises performing at least one process selected from the group consisting of:
structure detection; and
nodule detection.
5. The method according to claim 1 , wherein said processing at least one feature using decision fusion comprises at least one process selected from the group consisting of:
rib-crossing identification;
calcification identification;
obscured nodule identification; and
blood vessel identification.
6. The method according to claim 1 , further comprising:
marking said final detection results on a standard radiographic image.
7. The method according to claim 6 , wherein said marking comprises:
using different types of markings to represent different types of findings.
8. The method according to claim 6 , wherein a display of said markings enables a user to obtain further information regarding a final detection result associated with a given marking.
9. The method according to claim 8 , wherein said further information comprises at least one item selected from the group consisting of: a calcification score, a region of interest, and a result type.
10. The method according to claim 1 , wherein said processing at least one of said derived images includes at least one portion that is adapted to be performed using parallel processing of more than one of said derived images.
11. A machine-readable medium containing software code that, when executed by a processor, causes the processor to implement the method according to claim 1 .
12. A system for performing detection based on radiographic images, the method comprising:
at least one processor; and
a machine-readable medium containing software code that, when executed by the at least one processor, causes the at least one processor to implement the method according to claim 1 .
13. The system according to claim 12 , wherein said software code is downloaded from a remote medium to said machine-readable medium.
14. A method for performing detection based on radiographic images, the method comprising:
downloading from a remote medium software code that, when executed by a processor, causes the processor to implement the method according to claim 1 ; and
executing said software code on the processor.
15. A system for performing detection based on radiographic images, the system comprising:
a first processing module to process at least two radiographic images obtained using different energy levels to obtain derived images at least a soft-tissue equivalent image, a bone equivalent image, and an image including both soft tissue and bone;
a second processing module to process at least one of said derived images to detect at least one feature within said at least one derived image, wherein said second processing module is configured to perform at least two processes selected from the group consisting of multi-resolution enhancement, object feature extraction, linear neural network classification, and fuzzy neural network classification, and wherein said second processing module further comprises a decision fusion module to process results obtained from said at least two processes; and
a decision fusion module to process said at least one feature using decision fusion to obtain one or more final detection results.
16. The method according to claim 15 , wherein said decision fusion module to process said at least one feature is adapted to perform at least one process selected from the group consisting of:
rib-crossing identification;
calcification identification;
obscured nodule identification; and
blood vessel identification.