Semi-Automated Image Registration for Soft Tissue Biopsy Guidance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for registering real-time ultrasound images with pre-acquired diagnostic images, such as MR or CT images, face challenges due to differences in resolution, clarity, and anatomical markers, leading to unreliable automated registration, especially in soft tissue biopsies like prostate cancer diagnosis, where TRUS-guided biopsies often fail to detect cancer correctly.
Innovation Solution
An apparatus and method that use a localizer and registration unit to determine baseline and motion correction transforms for registering 3D diagnostic and ultrasound images, employing iterative automated registration with different similarity measures in each dimension, and a user interface for review and adjustment, allowing for accurate semi-automatic registration independent of tracking systems and fiducials.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated registration is used to align ultrasound and diagnostic images, then registration speed is improved, but reliability deteriorates due to differences in image resolution, clarity, and anatomical markers
Solution Approach 1:
The patent introduces a semi-automated registration process where a computer-generated composite image serves as an intermediary between the ultrasound image and the diagnostic image. The system presents multiple candidate transformations to the operator, who selects the best match. This intermediary step allows automated computation of multiple options while maintaining reliability through human verification, directly resolving the contradiction between speed and accuracy in image registration.
2Measurement precision
If fiducials and tracking systems are used for image registration, then measurement precision is improved, but device complexity and ease of operation worsen
Solution Approach 1:
The patent extracts and eliminates the need for fiducials and external tracking systems from the registration process. Instead of requiring these additional physical components, the system uses image-based features and computer-generated composite images to achieve registration. This removes the complexity of fiducial placement and tracking system integration while maintaining measurement precision through the semi-automated image matching process.
3Reliability
If manual registration is performed by operators, then reliability is improved through visual verification, but productivity deteriorates due to time-consuming procedures
Solution Approach 1:
The patent segments the registration task into two distinct phases: an automated phase that generates multiple candidate transformations quickly, and a manual verification phase where the operator reviews and selects the best match. This segmentation allows the system to leverage the speed of automated computation for generating options while preserving the reliability of human judgment for final selection, thereby improving overall productivity without sacrificing reliability.
Data Source
AI summary
A 3D ultrasound image from a memory (20) is compared with a 3D diagnostic image from a memory (12) by a localizer and registration unit (30) which determines a baseline transform (Tbase) which registers the 3D diagnostic and ultrasound volume images. The target region continues to be examined by an ultrasound scanner (22) which generates a series of real-time 2D or 3D ultrasound or other lower resolution images. The localizer and registration unit (30) compares one or a group of the 2D ultrasound images with the 3D ultrasound image to determine a motion correction transform (Tmotion). An image adjustment processor or program (32) operates on the 3D diagnostic volume image with the baseline transform (Tbase) and the motion correction transform (Tmotion) to generate a motion corrected image that is displayed on an appropriate display (74).