AI Video Frame Aggregation for Endoscopic Tissue Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current endoscopic technologies face challenges in visualizing the entire internal surface of body cavities during procedures like colonoscopy, leading to missed polyps or cancers due to difficulty in identifying and stitching together partial images from different frames, which limits the effectiveness of AI analysis and adenoma detection rates.
Innovation Solution
An AI system that recognizes and groups portions of an area of interest across multiple video frames in real-time, providing a comprehensive data picture by stitching together subareas and guiding the endoscopist to collect additional information for enhanced image analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional endoscopic imaging is used to visualize the entire internal surface of the colon, then the procedure can be performed with standard equipment, but polyps or cancers are missed due to difficulty in identifying and stitching together partial images from different frames
Solution Approach 1:
The patent divides the colon visualization task into multiple video frames captured from different positions and angles. Each frame captures a partial view of the tissue, and the AI system segments and analyzes these individual frames to identify polyps or abnormalities, then stitches the results together to form a complete visualization of the entire colon surface.
Solution Approach 2:
The AI system provides real-time feedback to the endoscopist by analyzing video frames as they are captured and indicating whether additional frames are needed from specific areas. This feedback loop ensures that the system collects sufficient data to achieve complete tissue visualization and accurate polyp detection, addressing the information loss problem.
2Measurement precision
If multiple video frames are captured to ensure complete visualization of the colon mucosa, then detection accuracy improves, but the complexity of stitching and analyzing the frames increases
Solution Approach 1:
The patent introduces an AI system as an intermediary between the endoscopist and the video frames. This intermediary automatically performs the complex tasks of frame analysis, polyp detection, and stitching without requiring the endoscopist to manually process each frame. The AI system handles the computational complexity while providing simplified guidance to the operator.
Solution Approach 2:
The patent replaces manual image stitching and analysis with an automated AI-based system. Instead of relying on mechanical or manual methods to combine and analyze multiple video frames, the system uses machine learning algorithms to automatically detect features, track tissue across frames, and synthesize a complete visualization, significantly reducing operational complexity.
3Ease of manufacture
If the endoscopist manually ensures complete visualization of the entire internal surface, then no additional technology is needed, but it is extremely challenging to assure complete visualization and polyps are missed
Solution Approach 1:
The patent enables the system to perform self-service by automatically tracking which areas of the colon have been visualized and identifying gaps in coverage. The AI system monitors the video frames, determines whether complete visualization has been achieved, and prompts the endoscopist to examine specific areas that have been missed, ensuring reliable and complete examination without requiring manual tracking by the operator.
Solution Approach 2:
The patent replaces the manual cognitive task of tracking and integrating images mentally with an automated computer-based system. The AI systematically processes video frames, maintains a record of visualized areas, and ensures complete coverage, significantly improving reliability while keeping the procedure simple for the endoscopist to follow.
Data Source
AI summary
Methods and systems are provided for aggregating features in multiple video frames to enhance tissue abnormality detection algorithms, wherein a first detection algorithm identifies an abnormality and aggregates adjacent video frames to create a more complete image for analysis by an artificial intelligence detection algorithm, the aggregation occurring in real time as the medical procedure is being performed.