Surgical Bleeding Analysis From Endoscopic Images Using AI
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional endoscopic images are inadequate for accurately evaluating internal bleeding in patients, lacking sufficient information for surgeons to make informed decisions about appropriate procedures.
Innovation Solution
A computing device utilizes deep learning models and optical flow tracking to generate surgical condition information, including bleeding warnings, point identifications, flow path indications, and amount/speed indications, based on surgical image sets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If conventional endoscopic images are used to evaluate internal bleeding, then the imaging process is simple and quick, but the information provided is insufficient for making informed surgical decisions
Solution Approach 1:
The patent introduces an AI analysis system as an intermediary between the endoscopic imaging system and the surgeon. This intermediary processes raw endoscopic images through deep learning models to extract meaningful bleeding information, effectively bridging the gap between simple imaging and comprehensive decision-making support without requiring the surgeon to directly analyze complex raw images
Solution Approach 2:
The patent replaces the manual mechanical analysis process (surgeon visually examining endoscopic images) with an automated AI-based image recognition system. The deep learning models automatically detect, classify, and quantify bleeding conditions, substituting human visual assessment with algorithmic analysis that provides more consistent and detailed information
2Measurement precision
If multiple deep learning models and optical flow tracking are implemented, then bleeding detection precision is improved, but processing time and computational resources increase
Solution Approach 1:
The patent employs multiple specialized deep learning models that have been pre-trained for specific bleeding detection tasks. These models are prepared in advance with pre-defined detection algorithms and parameters, allowing them to quickly process images without requiring complex real-time calculations, thus maintaining high precision while reducing processing time
Solution Approach 2:
The patent divides the bleeding detection task into multiple specialized sub-tasks handled by different deep learning models (e.g., bleeding point detection, bleeding area measurement, blood flow direction analysis). Each model focuses on a specific aspect of bleeding detection, improving overall precision while allowing parallel processing that reduces total computation time
Data Source
AI summary
The present application provides a method of generating surgical condition information. A surgical condition information is generated based on a surgical image set that is associated with a surgical site by a computing device. The method includes receiving the surgical image set and automatically generating the surgical condition information based on the surgical image set. The surgical condition information contains a bleeding warning, a bleeding point identification, a blood flow path indication, a prompting bleeding point, a bleeding amount indication, or a bleeding speed indication. In addition, a computing device of generating surgical condition information is also provided in the present application.