Segmentation and Fracture Detection in CT Images
AssigneeVirginia Commonweath University
InventorsJie Wu, Rosalyn Hobson Hargraves, Kayvan Najarian, Ashwin Belle, Kevin R. Ward
Life of this patent
Estimating the term…
Abstract
Provided is a new hierarchical methodology having a series of computational steps such as adaptive window creation, 2-D SWT application, masking, and boundary tracing is proposed. The techniques and systems are able to detect and quantify fracture as well as to generate recommendations for decision-making and treatment planning in traumatic pelvic injuries.
Claims23
A method for analysis of computed tomography (CT) images for diagnosis of bone fractures, the method implemented in a decision support system having a processor and a memory, the method comprising:
- obtaining a CT image of a bone and performing a pre-processing on the obtained CT image to remove a first set of features to created a pre-processed CT image;
- automatically performing an edge detection on the pre-processed CT image to form an edge enhanced CT image and comparing the edge enhanced CT image to a set of bone image templates;
- automatically identifying from among the bone image templates a matching bone image template corresponding to the edge enhanced CT image, wherein the bone image template is a best matching template identified using an affine transformation dependent correlation between the set of bone image templates and the edge enhanced CT image;
- automatically performing segmentation on the edge enhanced CT image using the matching bone template using an Active Shape Model (ASM) technique, wherein the edge enhanced CT image and the matching bone template are registered to one another forming a registered ASM technique;
- and
- detecting a fracture in the edge enhanced CT image in response to application of the registered ASM technique.
An apparatus comprising:
- a computer processor;
- and
- a memory storing computer-readable instructions that, when executed by the computer processor, cause the computer processor to,
- obtain a CT image of a bone and perform a pre-processing on the obtained CT image to remove a first set of features to created a pre-processed CT image;
- perform an edge detection on the pre-processed CT image to form an edge enhanced CT image and compare the edge enhanced CT image to a set of bone image templates;
- identify from among the bone image templates a matching bone image template corresponding to the edge enhanced CT image, wherein the bone image template is a best matching template identified using an affine transformation dependent correlation between the set of bone image templates and the edge enhanced CT image;
- perform segmentation on the edge enhanced CT image using the matching bone template using an Active Shape Model (ASM) technique, wherein the edge enhanced CT image and the matching bone template are registered to one another forming a registered ASM technique;
- and
- detect a fracture in the edge enhanced CT image in response to application of the registered ASM technique.
A method for analysis of computed tomography (CT) images for diagnosis of bone fractures, the method implemented in a decision support system having a processor and a memory, the method comprising:
A method for analysis of computed tomography (CT) images of bone images, the method implemented in a decision support system having a processor and a memory, the method comprising:
- a) obtaining a plurality of bone image templates, each bone image template obtained for a different image slicing of a bone region of a body;
- b) obtaining one or more training images of the bone region, the one or more training images taken from a different source than the bone template images;
- c) correlating the one or more training images to the plurality of bone image templates to identify a matching bone template, wherein the matching bone template is a determined from an optimization;
- d) if the matching bone template is not identified from the optimization, then performing an affine transformation on the plurality of bone image templates and/or the one or more training images and performing c) again;
- e) if the matching bone template is identified, then storing a registered image of the matching bone template and/or of the one or more training images.
An apparatus comprising:
- a computer processor;
- and
- a memory storing computer-readable instructions that, when executed by the computer processor, cause the computer processor to,
- a) obtain a plurality of bone image templates, each bone image template obtained for a different image slicing of a bone region of a body;
- b) obtain one or more training images of the bone region, the one or more training images taken from a different source than the bone template images;
- c) correlate the one or more training images to the plurality of bone image templates to identify a matching bone template, wherein the matching bone template is a determined from an optimization;
- d) if the matching bone template is not identified from the optimization, then perform an affine transformation on the plurality of bone image templates and/or on the one or more training images and perform c) again;
- e) if the matching bone template is identified, then store a registered image of the matching bone template and/or of the one or more training images.
Dependencies are inferred from the claim text.