Ultrasound Probe Pose Estimation for Cardiac Image Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for real-time fusion of 2D and 3D images with 2D fluoroscopic images in cardiac interventions are challenging due to the lack of synchronization and alignment of soft tissue visualization, requiring additional hardware and being computationally expensive, which can disrupt clinical workflows and lead to inaccuracies.
Innovation Solution
A robust learning-based method that automatically extracts patient-specific models from non-contrasted 3D C-arm CT images to align pre-operative anatomical information with live 2D fluoroscopy, using sparse matching and probabilistic models to estimate the pose of the ultrasound probe in 6 degrees of freedom without additional hardware, allowing for visualization of soft tissue anatomy without contrast agents or rapid pacing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If hardware-based tracking devices are used to align ultrasound and fluoroscopy coordinate systems, then image registration accuracy is improved, but device complexity and operational disruption increase
Solution Approach 1:
The patent replaces hardware-based electromagnetic or mechanical tracking devices with an image-based pose estimation method. The system detects the ultrasound probe in fluoroscopic images using computer vision algorithms, eliminating the need for additional tracking hardware while achieving accurate registration through software-based pose detection and transformation.
Solution Approach 2:
The patent creates a virtual copy of the ultrasound probe's position and orientation by detecting visual features in fluoroscopic images. Instead of using physical trackers, the system generates a digital representation of the probe pose through image processing, allowing coordinate system alignment without additional hardware.
2Measurement precision
If additional tracking hardware is introduced for image fusion, then registration accuracy is improved, but clinical workflow disruption increases
Solution Approach 1:
The patent substitutes mechanical tracking hardware with automated image-based pose estimation. The system processes existing fluoroscopic images to detect probe position and orientation, eliminating the need for additional devices that would disrupt the sterile field and clinical workflow while maintaining registration accuracy.
Solution Approach 2:
The system uses the existing fluoroscopic images themselves to provide the registration information, making the imaging system self-sufficient. The fluoroscopy system detects the ultrasound probe and provides transformation data without requiring external tracking devices, integrating the function into the existing workflow.
3Productivity
If rigid transformation is used for ultrasound image registration, then processing speed is improved, but adaptability to probe rotation and zoom changes deteriorates
Solution Approach 1:
The patent transitions from rigid transformation to affine transformation, allowing the registration model to dynamically adapt to probe rotation, zoom, and other geometric changes. The affine transformation includes additional parameters for rotation, scaling, and shearing, enabling the system to handle non-rigid geometric variations while maintaining computational efficiency through linear algebra operations.
4Measurement precision
If 3D to 2D image registration is performed in real-time, then navigation accuracy is improved, but computational cost increases
Solution Approach 1:
The patent extracts only the essential transformation parameters (6 degrees of freedom: 3 translations and 3 rotations) from the complex 3D to 2D registration problem. By focusing on the pose estimation of the ultrasound probe and using efficient image-based detection algorithms, the system achieves accurate navigation while minimizing computational requirements through targeted processing.
Solution Approach 2:
The patent replaces computationally intensive 3D model-based registration with a 2D image-based pose estimation approach. By detecting the ultrasound probe in 2D fluoroscopic images and calculating transformation parameters directly from image data, the system reduces computational complexity while maintaining navigation accuracy.
Data Source
AI summary
A method for real-time fusion of a 2D cardiac ultrasound image with a 2D cardiac fluoroscopic image includes acquiring real time synchronized US and fluoroscopic images, detecting a surface contour of an aortic valve in the 2D cardiac ultrasound (US) image relative to an US probe, detecting a pose of the US probe in the 2D cardiac fluoroscopic image, and using pose parameters of the US probe to transform the surface contour of the aortic valve from the 2D cardiac US image to the 2D cardiac fluoroscopic image.