Method, computer program product and computer system for correcting a CT image
Inventors
ANDERSSON, Sebastian • NILSSON, Rasmus • TRANEUS, Erik
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-11055884-B2
Publication Date
2021-07-06
Expiration Date
Abstract
A method of correcting a first CT image for shading artefacts using another CT image as reference, comprising the steps of creating a difference image between the images and a correction map based on the difference image. The correction map is used to correct the first CT image. A method of converting a CT image to the value range of another CT image is also disclosed.
Core Innovation
The invention corrects artefacts in a first CT image of a region of a patient by registering the first CT image with a corresponding second CT image and ensuring the first and second CT images use the same value range. From the value matched image pair, the invention obtains a difference image D and defines an image mask M for voxels to be considered. This artefact-focused difference information is then used to derive a correction map for removing shading artefacts.
A final correction map Cf is obtained by an iterative method that defines and updates a correction map C based on the difference image and optionally the image mask M. The method updates the final correction map Cf and updates the first CT image using the correction map, taking the updated image as the current first CT image. The image mask M is updated using information from the correction map, the difference image and/or the current first and second CT image, and difference images between updated images are recalculated to support further updates.
After convergence of the iterative process, the invention applies the final correction map to the original first CT image or to the current first CT image. The document further describes obtaining data representing the registration between the images, and includes a value range conversion approach based on converting images to a common value range using a transform representation derived from deformable registration and joint histogram analysis.
Claims Coverage
The independent claim is a method that combines registration, value range matching, difference-image computation, artefact voxel selection via an image mask, and an iterative derivation of a final correction map with application to the first CT image. Across the family, the independent claim is refined by additional inventive features, including how the value-range matching is implemented using transform representations derived from joint histograms and tissue/material value pairs, and how the image mask is defined using threshold-based rules on voxel-wise absolute differences.
Registration-based difference computation with value range matching
Obtaining the first CT image and a corresponding second CT image, obtaining data representing a registration between the images, ensuring that the first and the second CT image use the same value range to form a value matched image pair, and obtaining a difference image D between the images of the value matched image pair.
Image mask defining voxels for artefact-focused correction
Defining an image mask M defining which voxels to consider in the difference image.
Iterative final correction map derivation and application
Obtaining a final correction map Cf by an iterative method comprising defining a correction map C based on the difference image and optionally the image mask M, updating the final correction map Cf, updating the first CT image using the correction map, updating the image mask M based on information from the correction map, the difference image and/or the current first and second CT image, updating the first or the second CT image of the value matched image pair based on the current first CT image, calculating a difference image between the updated second CT image and the current first CT image, and applying the final correction map to the original first CT image or to the current first CT image.
Value-range conversion using transform representation from joint histogram tissue/material pairs
Obtaining a transform representation by providing and registering first and second CT images, creating a joint histogram from the registration, identifying at least two tissue value pairs from the joint histogram, and using those pairs to create a transformation representation between the two CT images in one or both directions.
Threshold rule for image mask selection using absolute difference
Defining an image mask that includes voxels whose absolute difference is below a threshold and excludes voxels whose difference is above the threshold.
Transform representation enabling value conversion in one or both directions
Using a transform representation between the two CT images in one or both directions as part of value range matching.
Computer-executable implementation as a non-transitory program product
Providing a computer program product comprising non-transitory computer-readable code that, when executed, causes a computer to perform the method of claim 1.
Across the claims, the core inventive approach is to register a second CT to a first CT, match value ranges to form a value matched image pair, compute a difference image, and use an artefact voxel mask in an iterative procedure that updates a correction map until a final correction map is obtained and applied to the first CT image. Dependent claim features further specify value-range conversion using transform representations derived from joint histogram tissue/material value pairs and specify threshold-based mask selection rules, with additional dependent features covering software implementation.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Not explicitly described in patent.
Interested in licensing this patent?