In Vivo Image Stream Segmentation for Reduced Review Time
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current in-vivo imaging systems face challenges in efficiently processing and analyzing large volumes of image data from the gastrointestinal tract, requiring lengthy review times for physicians due to the vast number of images captured during procedures.
Innovation Solution
A computer-implemented method and system for segmenting an image stream into multiple segments by calculating pixel-based properties, identifying points of change, and displaying a summarized representation using a color bar, which aids in reducing analysis time and improving diagnostic efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If physicians review all captured images manually, then diagnostic accuracy is maintained, but review time becomes excessively long
Solution Approach 1:
The patent segments the image stream into multiple segments based on detected change points in pixel-based properties. This segmentation allows physicians to review representative frames from each segment rather than all individual frames, significantly reducing review time while maintaining diagnostic accuracy through strategic sampling of key segments.
Solution Approach 2:
The system performs preliminary automated analysis of the image stream to identify change points and select representative frames before physician review. This preliminary action filters out redundant frames and highlights clinically significant segments, enabling physicians to focus their attention on critical areas and reduce overall review time.
2Measurement precision
If frame rate is increased to capture more detailed motility events, then measurement precision improves, but data volume and processing complexity increase
Solution Approach 1:
The patent extracts only the essential pixel-based properties (mean intensity, standard deviation, skewness, kurtosis) from the full image data at each frame. This extraction approach maintains the ability to detect motility events with high precision while significantly reducing the data volume and processing complexity compared to analyzing complete high-resolution images at every frame.
Solution Approach 2:
The system changes the parameter representation from full image data to simplified statistical descriptors (mean, standard deviation, skewness, kurtosis) of pixel intensities. This parameter transformation enables efficient detection of motility events by capturing essential information in a compressed form that reduces computational burden while maintaining diagnostic precision.
3Loss of information
If all image frames are displayed individually, then complete information is preserved, but information overload occurs and analysis efficiency decreases
Solution Approach 1:
The patent applies partial action by displaying only a selected subset of frames (representative frames from each segment) rather than all frames. This selective display maintains sufficient information for accurate diagnosis while eliminating redundant information, thereby improving analysis efficiency without causing information overload.
Solution Approach 2:
The system adds a temporal organization dimension to the display by grouping frames into segments and selecting representative frames from each segment. This dimensional reorganization transforms the overwhelming sequence of individual frames into a structured overview with hierarchical organization, enabling physicians to grasp essential information quickly while maintaining access to complete data when needed.
Data Source
AI summary
A system and method for segmenting an image stream to a plurality of segments is provided. A processing unit may be configured to calculate a set of pixel-based properties for frames of an image stream, and detect segments of constant mean values in the sets of pixel-based properties. The segments of constant mean values are detected by using a window method that determines possible partition points of the window, and calculating a difference between mean values of the sets of pixel-based properties for each sub-window of frames. Points of change may be identified in the calculated difference between mean values of the sets of pixel-based properties. A display unit may display a summarized representation of the image stream, wherein the summarized representation includes a plurality of segments, each segment corresponding to a segment of constant mean values in the sets of pixel-based properties.