Multi-Scale Registration for Dual Energy Image Misregistration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Dual energy imaging faces misregistration errors due to anatomical motion between high- and low-energy image acquisitions, leading to artifacts in combined images, particularly caused by bulk, cardiac, and respiratory motions, which existing methods fail to adequately address across various scales and resolutions.
Innovation Solution
A multi-resolution, multi-scale image registration method that selects regions of interest of varying dimensions, uses a registration metric like mutual information to maximize alignment between high- and low-energy images, and interpolates transformations to correct misregistration errors, reducing computation complexity through downsampling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If double-shot acquisition is used to improve detective quantum efficiency and detectability index, then image quality is improved, but misregistration between high- and low-energy images occurs due to anatomical motion during the time lapse
Solution Approach 1:
The image registration process is segmented into multiple resolutions and scales. The patent divides the registration task into coarse-to-fine stages, where each stage operates at a different resolution level. This segmentation allows the system to address different types of motion (bulk vs. fine-scale) at appropriate scales, resolving the contradiction between maintaining registration accuracy and accommodating anatomical motion during double-shot acquisition.
Solution Approach 2:
The patent introduces a multi-resolution dimension to the registration problem. By processing images at multiple scales (from downsampled coarse versions to full-resolution fine details), the system adds a resolution dimension to the registration process. This enables simultaneous correction of both large-scale bulk motion and fine-scale anatomical motion, maintaining registration accuracy while preserving the benefits of double-shot acquisition.
2Manufacturing precision
If multi-scale registration is performed to correct misregistration errors at various anatomical scales, then registration accuracy is improved, but computation complexity increases
Solution Approach 1:
The patent applies preliminary downscaling (decimation) to create lower-resolution versions of the images before performing registration. This preliminary action reduces the computational burden of subsequent processing steps. By first establishing coarse alignment at downsampled resolutions and then progressively refining at higher resolutions, the system achieves multi-scale registration accuracy while keeping computation manageable through the preliminary reduction step.
Solution Approach 2:
The registration process uses dynamic resolution adjustment, transitioning from coarse to fine scales adaptively. The system dynamically selects which resolution level to process based on the current registration state and computational resources. This dynamic approach allows the algorithm to spend more computational effort where needed (fine-scale corrections) while using fewer resources for coarse adjustments, balancing accuracy and complexity.
Data Source
AI summary
A method is disclosed for resolving misregistration errors resulting from dual energy double-shot projection radiographic image acquisition. The method involves an iterative multi-scale, multi-resolution registration process that corrects misregistration errors progressively at scales ranging from bulk anatomical drift down to smaller scale motion such as that of fine pulmonary vasculature. The method may be incorporated as part of a dual energy image processing chain to create dual energy images with improved image quality and diagnostic performance.