Multi-Stage Pathological Image Model for Annotation Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The labor-intensive and time-consuming process of annotating pathological images for machine learning training data hinders the improvement of diagnostic accuracy in medical imaging.
Innovation Solution
A determination support device employing a multi-stage trained model that facilitates the derivation of diagnostic estimation results by annotating and training on pathological images, utilizing a hierarchical structure and stepwise training approach to enhance annotation efficiency and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a large amount of training data is used to improve machine learning accuracy, then diagnostic accuracy is improved, but the labor and time required for annotation increases significantly
Solution Approach 1:
The annotation process is segmented into multiple stages with progressively finer granularity. First, rough annotation is performed at a coarse level to cover large areas quickly, then detailed annotation is performed on specific regions of interest. This segmentation allows annotators to efficiently create training data at multiple levels of detail, reducing overall annotation time while maintaining sufficient accuracy for machine learning training.
2Manufacturing precision
If detailed annotation is performed to improve training data quality, then machine learning accuracy is improved, but the complexity and time of the annotation process increases
Solution Approach 1:
The annotation process is divided into multiple stages with different levels of detail. Rough annotation covers entire regions at a coarse level, while detailed annotation focuses only on specific regions of interest at a fine level. This staged approach reduces annotation complexity by breaking down the task into manageable segments that can be performed systematically.
Solution Approach 2:
Different levels of annotation detail are applied to different regions of the pathological image based on their importance. High-priority regions receive detailed annotation with fine granularity, while low-priority regions receive rough annotation at a coarser level. This local quality approach ensures training data quality where it matters most while reducing overall annotation complexity and time.
Data Source
AI summary
Making it possible to improve accuracy of machine learning. A determination support device includes: a derivation unit (100) that derives an estimation result of determination on a first pathological image obtained by imaging a biological sample, the derivation being performed using a multi-stage trained model in which class classification can be set in each stage.