Image Processing Device for Capsule Endoscopy Sequence Summarization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods are inefficient in handling large sequences of in vivo images, such as those from capsule endoscopes, as they require extensive time to identify and summarize relevant features like lesions, especially when images undergo non-rigid deformations due to elastic movements within the body.
Innovation Solution
An image processing device and method that calculates deformation information between images using areas other than the feature area, allowing for accurate identicalness determination of feature areas across images by projecting and comparing deformation areas, thereby summarizing image sequences effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all images in a large image sequence are displayed for observation, then complete diagnostic information is obtained, but the time required for observation increases significantly
Solution Approach 1:
The patent extracts only the essential images from a large image sequence by detecting lesions and determining identicalness across frames. Instead of displaying all 60,000 images, the system extracts and displays only those images containing unique diagnostic information, significantly reducing observation time while maintaining diagnostic completeness
Solution Approach 2:
The patent creates a summarized representation of the image sequence by copying and displaying only representative images that contain unique lesion information. The identification information management unit maintains copies of image identification data, allowing the system to reference and display only necessary images rather than the entire sequence
2Productivity
If image processing detects and displays only images with detected lesions, then observation efficiency improves, but identical lesions are captured repeatedly across consecutive images requiring repeated observation
Solution Approach 1:
The patent implements a feedback mechanism where the system determines whether detected lesions are identical across consecutive images using image processing. When identicalness is detected, the system provides feedback to suppress display of redundant images, ensuring each unique lesion is observed only once while maintaining efficient observation workflow
Solution Approach 2:
The patent discards redundant image displays by identifying and eliminating consecutive images containing identical lesions. The system recovers and retains only the representative image from each sequence of identical lesion captures, preventing repeated observation of the same diagnostic feature while preserving all unique diagnostic information
3Device complexity
If image processing methods assume rigid object movement for identicalness determination, then processing simplicity is maintained, but accuracy deteriorates when non-rigid deformations occur in in vivo images
Solution Approach 1:
The patent transitions from static rigid-body assumptions to dynamic deformation-aware processing. The image processing unit calculates deformation information between images and uses this dynamic data to determine identicalness, allowing accurate tracking of lesions even when they undergo non-rigid deformations due to elastic movements within the body
Solution Approach 2:
The patent changes the processing parameters by incorporating deformation information calculations into the identicalness determination process. Instead of using fixed rigid-body transformation parameters, the system adapts its parameters to account for non-rigid deformations, improving measurement precision while managing processing complexity through efficient calculation methods
Data Source
AI summary
An image processing device includes a memory and a processor. The processor acquires an image sequence, calculates deformation information about images, performs a feature area detection process, performs an identicalness determination process on an ith feature area and an (i+1)th feature area based on the deformation information about an ith image and an (i+1)th image, and performs an image summarization process. The processor performs the deformation information calculation process that calculates the deformation information based on an ith deformation estimation target area that includes an area of the ith image other than the ith feature area, and an (i+1)th deformation estimation target area that includes an area of the (i+1)th image other than the (i+1)th feature area.