Model-Based Tomographic Reconstruction for Metal Artifact Reduction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Tomographic imaging of objects containing metallic components often suffers from reduced image quality due to metal streak artifacts, particularly in regions surrounding implants, where photon starvation and high attenuation lead to diagnostic challenges.

Innovation Solution

A model-based penalized-likelihood estimation approach that incorporates known component geometry and composition into the reconstruction algorithm, using an alternating maximization method to jointly estimate anatomy and component pose, thereby reducing artifacts and improving image quality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional tomographic reconstruction algorithms are used to image objects containing metallic components, then the reconstruction process is simple and fast, but metal streak artifacts appear and image quality deteriorates in regions surrounding implants

Engineering Contradiction:

Improveimage qualityVSAvoidreconstruction algorithm complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The reconstruction algorithm segments the image volume into two distinct components: a known component (metallic implant) and an unknown component (anatomical structures). By separating the imaging problem into these segments, the algorithm can apply different processing strategies to each, reducing artifacts while maintaining computational feasibility

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The algorithm performs preliminary actions by incorporating known information about the metallic component (geometry, material properties, position) into the reconstruction process before final image formation. This pre-processing step allows the algorithm to anticipate and compensate for photon starvation effects in regions surrounding the implant

Inventive Principle:

Principle #10Preliminary action

Solution Approach 3:

The algorithm changes parameters by using an unconstrained objective function that incorporates component models and imaging device models, allowing the reconstruction to optimize for reduced artifacts rather than simply following conventional reconstruction paths. This parameter change enables better image quality without prohibitively increasing complexity

Inventive Principle:

Principle #35Parameter changes

2Object-affected harmful factors

If measurements through metal are treated as missing data and eliminated from reconstruction, then artifact reduction is achieved, but information loss increases and diagnostic accuracy decreases

Engineering Contradiction:

Improvemetal streak artifactsVSAvoidprojection data

Core Design Contradiction:

Object-affected harmful factorsVSLoss of information

Solution Approach 1:

The algorithm converts the harmful effect of high attenuation by metal into a benefit by using the known component model to predict exactly how the metal affects the projections. This allows the algorithm to compensate for photon starvation rather than treating affected measurements as useless missing data, thereby reducing artifacts while preserving diagnostic information

Inventive Principle:

Principle #22Blessing in disguise (Convert harm into benefit)

Solution Approach 2:

The known component model acts as an intermediary between the raw projection data and the final reconstruction. It mediates the reconstruction process by providing information about the metallic component's effect on the projections, allowing the algorithm to correctly interpret measurements that would otherwise be treated as missing or corrupted data

Inventive Principle:

Principle #24Intermediary (Mediator)

3Measurement precision

If exact knowledge of component geometry and composition is incorporated into the reconstruction algorithm, then artifact reduction is improved, but computational complexity and processing time increase

Engineering Contradiction:

Improveartifact reductionVSAvoidreconstruction processing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The algorithm applies partial action by incorporating only the essential known information about the component (geometry, material properties, position) rather than requiring complete and perfect knowledge. This selective incorporation achieves sufficient artifact reduction without the computational burden of processing excessive or redundant information

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

The algorithm applies local quality by focusing computational effort on regions surrounding the metallic component where photon starvation effects are most pronounced. Rather than uniformly processing the entire image volume with high computational intensity, the algorithm concentrates resources where they are most needed for artifact reduction

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS9177374B2Model-based processing of image data

Publication Date: 2015.11.03 JOHNS HOPKINS UNIVERSITY

AI summary

An imaging system for processing image data of an object containing a component. The imaging system includes an imaging device arranged to obtain image data and a processor. The processor is adapted to receive the image data from the imaging device, obtain a component model for the component, obtain an imaging device model for the imaging device, construct an unconstrained objective function based on the component model and the imaging device model, and construct a model of the object containing the component based on the unconstrained objective function and the image data, and a display device adapted to display an image for the object containing the component based on the model.