Projection-to-Volumetric Image Registration With Guided Positioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Challenging to position a projection system during medical procedures to register intra-procedural projection images to pre-procedural volumetric images while providing sufficient detail of the region of interest, often requiring a time-consuming trial-and-error approach.
Innovation Solution
A computer-implemented method for registering medical images by receiving a volumetric image and a projection image, computing a spatial transformation based on identified landmarks, adjusting the projection system's position, and generating a second projection image that is robustly registered to the volumetric image, even without requiring additional landmarks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a zoomed-in projection image is used to provide detailed view of the region of interest, then the detail of the region of interest is improved, but the number of anatomical landmarks for registration is reduced
Solution Approach 1:
The system performs preliminary registration using a first projection image that captures a larger anatomical region with sufficient landmarks. The spatial transformation computed from this preliminary registration is then used to guide the projection system to a second position that provides a zoomed-in view of the region of interest, eliminating the need for landmarks in the final detailed view
Solution Approach 2:
The patent introduces an intermediary spatial transformation computation step that bridges the gap between landmark-based registration and landmark-poor detailed imaging. The transformation computed from landmark-matching in the first projection image serves as an intermediary that enables accurate registration of the second projection image even when it lacks sufficient anatomical landmarks
2Reliability
If a trial-and-error approach is used to position the projection system, then both registration and detail requirements can be achieved, but the time consumption increases
Solution Approach 1:
The system uses feedback from the first projection image registration to automatically determine the optimal positioning for the second projection image. The spatial transformation computed from landmark matching provides feedback that guides the projection system positioning, eliminating the need for trial-and-error adjustments by physicians
Solution Approach 2:
The patent replaces the manual mechanical trial-and-error positioning process with an automated computational system. Instead of physically moving the projection system through multiple trial positions, the system computes the optimal position mathematically based on the spatial transformation derived from landmark matching, substituting mechanical trial-and-error with computational determination
Data Source
AI summary
A computer-implemented method of registering medical images, is provided. The method includes computing a spatial transformation between a first projection image (P1) and a volumetric image. The first projection image (P1) is generated from a first position (POS1) of a projection imaging system (110). Input defining the region of interest (ROI) in the volumetric image is received. One or more adjustments (TP1→P2) to the first position (POS1) of the projection imaging system are then calculated based on the spatial transformation and the received input, to provide a second position (POS2) of the projection imaging system for generating a second projection image (P2) representing the region of interest (ROI). A second projection image (P2) representing the region of interest (ROI) is then received. The second projection image (P2) is generated from the second position (POS2). The second projection image (P2) is then registered to the volumetric image (V1) based on the spatial transformation and the calculated one or more adjustments (TP1→P2).