CNN Diagnostic Assistance for Endoscopic Image Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current endoscopic examinations for digestive organs, particularly the small bowel, face challenges in efficiently and accurately diagnosing conditions like H. pylori infections, colorectal diseases, and small bowel adenocarcinomas due to the high volume of images and subjective nature of human interpretation, leading to fatigue and potential inaccuracies.
Innovation Solution
A diagnostic assistance method and system utilizing a convolutional neural network (CNN) trained on endoscopic images to classify anatomical sites, detect H. pylori infections, and identify conditions such as erosion/ulcer in the small bowel, and determine the invasion depth of esophageal cancers, aiming to reduce the burden on specialists and improve diagnostic accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If endoscopic images are manually interpreted by specialists, then diagnostic accuracy can be maintained, but the time and effort required for image interpretation increases significantly
Solution Approach 1:
The patent introduces an intermediary system consisting of a preprocessing unit and a deep neural network that acts as a mediator between the endoscopic images and the specialist's final diagnosis. The preprocessing unit automatically detects candidate lesions and extracts features, preparing refined input for the specialist, thereby reducing manual interpretation time while preserving diagnostic accuracy.
Solution Approach 2:
The patent replaces the mechanical process of manual image scanning and initial assessment with an automated computer vision system. The deep neural network substitutes the human eye and brain's initial processing functions, automatically identifying potential lesions and calculating diagnostic probabilities, thus eliminating time-consuming manual preliminary review while maintaining accuracy through specialist verification.
2Quantity of substance
If the volume of endoscopic images to be reviewed increases, then more comprehensive examination coverage is achieved, but specialist fatigue and potential inaccuracies increase
Solution Approach 1:
The system introduces an intermediary automated analysis layer that processes large volumes of images before presenting them to specialists. The preprocessing unit identifies candidate lesions across all images and calculates diagnostic probabilities, allowing specialists to focus only on suspicious cases. This maintains diagnostic reliability by preserving specialist judgment for critical decisions while enabling comprehensive examination of large image sets without fatigue.
Solution Approach 2:
The patent applies partial action by having the automated system perform the preliminary screening and candidate identification tasks, reserving the specialist's full attention for only the most critical cases. The deep neural network performs excessive analysis by evaluating all images for potential lesions, then filtering results to present only relevant cases to specialists, thus maintaining reliability while managing workload.
3Productivity
If automated diagnostic systems are used, then processing speed increases, but diagnostic accuracy may decrease due to lack of human judgment
Solution Approach 1:
The patent merges the strengths of automated systems and human specialists into a hybrid diagnostic workflow. The preprocessing unit and deep neural network handle rapid image analysis and candidate identification, while specialists provide final diagnostic confirmation. This combination achieves high processing speed through automated initial screening while maintaining diagnostic accuracy through human judgment on critical cases.
Solution Approach 2:
The system uses the deep neural network as an intermediary that bridges automated processing and human judgment. The network processes images rapidly to generate candidate lesions and probability scores, then presents refined information to specialists for final diagnosis. This intermediary role enables fast processing while preserving human expertise for accuracy-critical decisions.
4Adaptability or versatility
If multiple diagnostic parameters are analyzed, then comprehensive disease characterization is achieved, but system complexity increases
Solution Approach 1:
The patent segments the diagnostic process into distinct functional modules: a preprocessing unit for image processing and candidate lesion detection, a deep neural network for feature extraction and probability calculation, and a specialist review interface. Each module handles specific diagnostic parameters independently, allowing comprehensive disease characterization through multiple parameters while managing system complexity through modular architecture and clear separation of concerns.
Data Source
AI summary
A diagnostic assistance method for a disease based on an endoscopic image of a digestive organ with use of a convolutional neural network (CNN). A diagnostic assistance method for a disease based on an endoscopic image of a digestive organ with a CNN trains the CNN using a first endoscopic image of the digestive organ and at least one final diagnosis result on positivity or negativity to the disease in the digestive organ, a past disease, a severity level, and information corresponding to a site where an image is captured, the final diagnosis result corresponding to the first endoscopic image, and the trained CNN outputs at least one of a probability of the positivity and/or the negativity to the disease, a probability of the past disease, a severity level of the disease, an invasion depth of the disease, and a probability corresponding to the site where the image is captured.