Pathology Slide Abnormal Region Training Using Synthetic Image Defects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing pathological slide image analysis algorithms face degradation due to errors, making it difficult to accurately extract or predict patient information, and manual error detection is challenging.
Innovation Solution
A method and system for training a machine learning model to detect abnormal regions in pathological slide images by generating training data using image processing techniques to create normal and abnormal regions, enabling the model to identify errors automatically.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual error detection is used on pathological slide images, then detection accuracy may be maintained, but the process becomes extremely time-consuming and difficult due to the rarity of errors
Solution Approach 1:
The patent applies preliminary action by pre-processing pathological slide images to enhance abnormal regions before detection. The system performs image preprocessing operations such as contrast enhancement and noise reduction on the entire image dataset beforehand, making subsequent error detection faster and more accurate without requiring manual examination of every image
Solution Approach 2:
The patent introduces an intermediary detection model that acts as a mediator between manual detection and final results. The model pre-identifies potential error regions, which are then verified or corrected through minimal human intervention, significantly reducing the time required compared to pure manual detection while maintaining high accuracy
2Productivity
If machine learning models are trained with insufficient abnormal region data, then training efficiency is maintained, but model performance degrades due to lack of diverse training examples
Solution Approach 1:
The patent applies copying by generating synthetic abnormal region data through image processing techniques. Normal pathological slide regions are copied and transformed with various abnormalities (noise addition, contrast changes, artifact insertion) to create diverse training examples, enabling the model to learn from abundant synthetic data while maintaining training efficiency
Solution Approach 2:
The patent uses parameter changes by systematically varying image parameters such as contrast, brightness, noise levels, and artifact types during synthetic data generation. This creates a wide range of training examples from limited original data, improving model reliability without requiring proportional increases in actual abnormal image collections
3Speed
If analysis algorithms process pathological slide images with errors, then processing speed is maintained, but information extraction accuracy degrades
Solution Approach 1:
The patent applies taking out by extracting and removing error regions from pathological slide images before analysis. The detection model identifies abnormal regions, which are then excised or masked out, allowing subsequent analysis algorithms to process only clean, error-free regions, thereby maintaining both processing speed and information extraction accuracy
Data Source
AI summary
A method, performed by at least one processor, for training a machine learning model for detecting an abnormal region in a pathological slide image is disclosed. The method including receiving one or more first pathological slide images, determining, from the received one or more first pathological slide images, a normal region based on an abnormality condition indicative of a condition of an abnormal region, generating a first set of training data including the determined normal region, generating the abnormal region by performing image processing corresponding to the abnormality condition with respect to at least partial region in the received one or more first pathological slide images, and generating a second set of training data including the generated abnormal region.