3D Image Registration via 2D Deformable Field Backprojection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current imaging systems, such as PET and CT, face significant image degradation due to respiratory motion, leading to inaccurate tumor localization and treatment planning, with existing motion compensation techniques being time-consuming and costly.
Innovation Solution
A method and system for registering 3D volumes by projecting them into 2D spaces, estimating 2D deformable fields, backprojecting these fields into 3D, and updating 3D deformable fields to accurately estimate and compensate for motion, reducing computational complexity and increasing accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If gated PET/CT technique with spline-based transformations or optical flow algorithms is used for motion estimation, then motion compensation accuracy is improved, but processing time and system complexity increase
Solution Approach 1:
The patent segments the 3D volume registration problem into multiple 2D slice registration problems. Each 2D slice is independently registered using deformable registration, and the resulting 2D deformation fields are then combined through backprojection to obtain the 3D deformation field. This segmentation reduces computational complexity while maintaining accuracy.
Solution Approach 2:
The patent transforms the complex 3D registration problem into simpler 2D registration problems by processing individual slices. The dimensionality reduction from 3D to 2D significantly decreases computational burden, and the results are synthesized back to 3D space through backprojection operations.
2Measurement precision
If gated PET/CT technique with advanced algorithms is used for motion estimation, then motion compensation accuracy is improved, but system cost increases due to additional circuitry
Solution Approach 1:
The patent segments the 3D volume registration problem into multiple 2D slice registration problems. Each 2D slice is independently registered using deformable registration, and the resulting 2D deformation fields are then combined through backprojection to obtain the 3D deformation field. This segmentation reduces computational complexity while maintaining accuracy.
Solution Approach 2:
The patent uses existing 2D deformable registration algorithms and techniques, which are well-established and can be implemented using standard computational resources. By copying and adapting 2D methods to solve the 3D problem through slice-by-slice processing, the system avoids the need for complex dedicated 3D hardware circuitry.
3Measurement precision
If active breath control is used to limit motion blurring, then image quality is improved, but patient comfort and feasibility deteriorate
Solution Approach 1:
The patent employs retrospective gating techniques that automatically analyze the acquired data to identify respiratory phases and sort events accordingly. The system self-adjusts by using the patient's natural breathing pattern without requiring active control, making the process comfortable and feasible for patients with serious illness.
Solution Approach 2:
The patent performs motion estimation and compensation during the image reconstruction process itself, rather than requiring pre-synchronization of breath holding. By preliminarily acquiring all necessary data and then retrospectively sorting and registering it according to respiratory phase, the system eliminates the need for active patient participation during critical imaging periods.
Data Source
AI summary
A method and system for registering two images is described. The method comprises synthesizing projections from two volumetric images to be registered, estimating a plurality of two dimensional (2D) deformable fields from the projections and generating a three dimensional (3D) deformable fields using a plurality of backprojections of the 2D deformable fields.