CT-Fluoroscopy Registration Using Segmented Vertebrae Inputs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Image registration in medical imaging is hindered by noisy and stacked anatomical elements in CT and fluoroscopy images, particularly in patients with high BMI or complex anatomy, leading to increased registration time and reduced accuracy due to overlapping gradients.
Innovation Solution
Segmentation of vertebrae and overlapping anatomical elements in CT and fluoroscopy images, followed by removal of noise and extraneous portions to facilitate accurate registration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image registration is performed on raw CT and fluoroscopy images with overlapping anatomical elements, then registration can be performed, but registration accuracy deteriorates due to noisy and stacked anatomical elements
Solution Approach 1:
The patent segments overlapping anatomical elements in CT and fluoroscopy images to separate vertebrae from other anatomical structures. This segmentation process creates clean, distinct representations of vertebrae that can be accurately registered without interference from overlapping ribs or other structures, directly resolving the accuracy deterioration caused by stacked anatomical elements
Solution Approach 2:
The patent extracts and removes noisy portions and extraneous elements from the images during the segmentation process. By taking out the harmful overlapping anatomical elements and noise, the system retains only the relevant vertebrae information needed for accurate registration, eliminating the harmful factors that degrade measurement precision
2Productivity
If image registration is performed on raw images with overlapping gradients, then registration can be performed, but registration time increases due to processing complexity
Solution Approach 1:
The patent performs segmentation and noise removal as preliminary actions before the actual registration process. By pre-processing the images to separate and clean the anatomical elements, the system reduces the complexity of the subsequent registration algorithm, allowing for faster processing without sacrificing accuracy
Solution Approach 2:
Segmentation simplifies the registration process by dividing complex overlapping images into distinct, manageable anatomical regions. This segmentation reduces the computational burden on the registration algorithm by eliminating the need to process and disambiguate overlapping gradients, thereby reducing registration time
3Measurement precision
If segmentation and noise removal are applied to images, then registration accuracy improves, but image processing complexity increases
Solution Approach 1:
The patent employs segmentation algorithms that automatically distinguish vertebrae from other anatomical elements based on geometric and intensity characteristics. This automated segmentation reduces the need for manual intervention and complex post-processing, maintaining high accuracy while managing processing complexity through algorithmic efficiency
Solution Approach 2:
By extracting and removing only the necessary noisy portions and extraneous elements, the patent minimizes the amount of processing required. The extraction process targets specific harmful elements rather than processing the entire image, reducing overall processing complexity while maintaining the accuracy benefits of clean data
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A system according to embodiments of the present disclosure comprises: a processor; and a memory storing instructions that, when executed by the processor, cause the processor to: receive a three-dimensional (3D) image of a patient's anatomy; segment a first set of anatomical elements from the 3D image; cause an imaging device to capture one or more two-dimensional (2D) images of the patient's anatomy; segment a second set of anatomical elements from the one or more 2D images; clean the one or more 2D images by removing at least one gradient line from each 2D image of the one or more 2D images; and register the one or more cleaned 2D images to the 3D image based on the segmented first set of anatomical elements and the segmented second set of anatomical elements, wherein the segmenting of the second set of anatomical elements further comprises defining a boundary around at least one anatomical tissue, wherein the boundary defines an area indicating an overlap between the at least one anatomical tissue and an anatomical object of the segmented second set of anatomical elements.