Endoscope Positioning via Depth Map Image Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Endoscopic image matching with virtual endoscopic images is challenging due to noise from body fluids or imaging limitations, making it difficult to accurately navigate the endoscope to a target position within tubular structures like bronchi, especially in multi-branched structures.
Innovation Solution
An examination support device that acquires and converts actual and virtual endoscopic images into depth images, allowing for similarity calculation and position estimation of the endoscope within the tubular structure, even in the presence of noise, by converting image expression forms into a common format and using pixel value differences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If virtual endoscopic images are converted into gray scale for matching, then the matching process becomes simpler, but noise from body fluids or imaging limitations cannot be effectively removed, making accurate position specification impossible
Solution Approach 1:
The patent introduces depth information as an intermediary element between the actual endoscopic image and the virtual endoscopic image. By converting both images into depth maps and comparing depth information rather than directly comparing the noisy visual images, the system achieves both simple processing and high accuracy. The depth map serves as a mediator that filters out noise from body fluids while preserving structural information for accurate position specification.
2Device complexity
If conventional image matching methods are used, then the processing is straightforward, but noise from body fluids or objects not captured in tomographic imaging prevents accurate matching and position specification
Solution Approach 1:
The patent extracts depth information from both actual and virtual endoscopic images, separating the structural geometric information from the noisy visual information. By taking out only the depth map data for comparison purposes, the system eliminates the influence of body fluids, mucus, and other noise elements that are present in the visual images but irrelevant for position determination. This extraction approach maintains processing simplicity while significantly improving matching reliability.
3Loss of information
If gray scale conversion is applied to virtual endoscopic images, then color and texture information is simplified, but noise from clouded lenses or unimaged objects remains, preventing accurate distal end position determination
Solution Approach 1:
The patent changes the parameter used for image comparison from visual intensity (color and texture) to depth information. By transforming both actual and virtual endoscopic images into depth maps, the system fundamentally changes the comparison parameter from surface appearance to spatial structure. This parameter change automatically filters out noise related to color, texture, and visual obstructions while preserving the geometric information necessary for accurate position determination.
Data Source
AI summary
An image acquisition unit acquires a first medical image in a first expression form and a second medical image in a second expression form different from the first expression form. In a case where a conversion unit converts the first expression form of the first medical image and the second expression form of the second medical image into a third expression form, a converted first medical image and a converted second medical image are acquired. A similarity calculation unit calculates similarity between the converted first medical image and the converted second medical image.