Multiplanar Reformation With Landmark-Guided Image Enhancement

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional multiplanar reconstruction (MPR) methods in medical imaging often result in image artifacts such as blur or step edges when reformating low or medium resolution image sequences, and imposing high-resolution constraints increases scan time and is prone to patient motion or contrast changes.

Innovation Solution

A method using a trained image enhancement network that accounts for the orientation of the reformatted image sequence relative to the originally acquired sequence, informed by a landmark plane, to attenuate artifacts and enhance resolution without requiring high-resolution image acquisition.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If high-resolution constraints are imposed on image sequences before reformatting, then image quality is improved, but scan time increases and the system becomes more susceptible to patient motion or contrast changes

Engineering Contradiction:

Improveimage qualityVSAvoidscan time

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

A deep learning-based image enhancement network is introduced as an intermediary between the reformatted image sequence and the final output. This network enhances the resolution of reformatted images derived from medium or low-resolution source data, achieving high-quality output without requiring high-resolution input scans, thereby avoiding increased scan time and susceptibility to motion artifacts

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system changes the resolution parameter of the output image through computational enhancement rather than through acquisition parameters. The deep learning network transforms medium or low-resolution input images into high-resolution output images by learning the mapping between different resolution levels, effectively decoupling output quality from input resolution requirements

Inventive Principle:

Principle #35Parameter changes

2Productivity

If retrospective MPR is performed on medium or low-resolution image sequences, then scan time is reduced and patient motion susceptibility is decreased, but image artifacts such as blur or step edges are introduced

Engineering Contradiction:

Improvescan efficiencyVSAvoidimage quality

Core Design Contradiction:

ProductivityVSManufacturing precision

Solution Approach 1:

The system converts the limitation of medium or low-resolution source data into an opportunity to demonstrate the capability of deep learning enhancement. By training the enhancement network specifically on pairs of low-resolution and corresponding high-resolution images, the system learns to compensate for the resolution limitations, turning what would be a source of artifacts into a training opportunity for artifact removal

Inventive Principle:

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

Solution Approach 2:

The patent replaces traditional mechanical/optical resolution enhancement methods (which would require high-resolution acquisition hardware and longer scan times) with a computational approach using deep learning networks. This substitution allows resolution enhancement to be achieved through software rather than hardware, maintaining scan efficiency while improving image quality

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Ease of operation

If conventional reformatting is performed without considering image orientation, then processing is simplified, but artifact attenuation is inconsistent across various planes of interest

Engineering Contradiction:

Improveprocessing simplicityVSAvoidartifact attenuation consistency

Core Design Contradiction:

Ease of operationVSReliability

Solution Approach 1:

The image enhancement network is trained to process different orientations of reformatted images with orientation-specific enhancement strategies. By providing the orientation information as input to the network, the system applies locally optimized enhancement parameters for each orientation, ensuring consistent artifact attenuation across axial, coronal, and sagittal planes while maintaining relatively simple processing through a unified network architecture

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS12499608B2Methods and systems for multiplanar reformation with machine learning based image enhancement

Publication Date: 2025.12.16 GE PRECISION HEALTHCARE LLC

AI summary

The disclosure relates to multiplanar reformation of three-dimensional medical images. In particular, the invention provides a method for reformatting image sequences by determining a landmark plane intersecting a volume, acquiring an image sequence, reformatting the image sequence along the landmark plane to produce a first reformatted image sequence, perturbing the landmark plane to produce a perturbed landmark plane, reformatting the first reformatted image sequence along the perturbed landmark plane to produce a second reformatted image sequence, mapping the second reformatted image sequence, the image sequence, and the landmark plane, to a resolution enhanced image sequence using a trained image enhancement network, and displaying the resolution enhanced image sequence via a display device. The present disclosure provides approaches which may reduce image artifacts in retrospectively reformatted image sequences, particularly in cases of retrospective reformatting of medium or low-resolution image sequences, without relying on acquisition of high-resolution 3D images.