Pathological Image Labeling with Partial-Image Overlays
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Generating large amounts of image-related learning data is burdensome for doctors, as they need to manually label numerous pathological images for training machine learning models, leading to a significant workload.
Innovation Solution
A data generation apparatus that displays a pathological image with superimposed borders of multiple partial images, allowing users to input labels for each partial image, and generates learning data by associating these images with their labels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If a large number of pathological images are manually labeled by doctors to generate learning data, then the quality and quantity of learning data improve, but the workload and time consumption for doctors increase significantly
Solution Approach 1:
The patent divides a large pathological image into multiple smaller partial images (tiles) and displays them in a grid layout overlaid on the original image. This segmentation allows doctors to label multiple regions simultaneously rather than processing one entire image at a time, significantly reducing the time required to generate learning data while maintaining data quality
2Measurement precision
If doctors manually label each pathological image individually, then accurate labels are obtained, but the process becomes extremely time-consuming and labor-intensive
Solution Approach 1:
By segmenting the labeling task into multiple partial images that can be processed in parallel, the system maintains labeling accuracy while dramatically improving productivity. The overlay display allows doctors to see the context of the entire image while focusing on specific regions, ensuring accurate labeling at scale
Solution Approach 2:
The patent combines multiple partial images into a single overlay display that shows the entire pathological image context. This merging allows doctors to maintain accurate labeling by seeing the global context while efficiently processing multiple regions simultaneously, thus improving both accuracy and productivity
Data Source
AI summary
There is provided a data generation apparatus including a display control unit that displays, on a screen, an image and borders of a plurality of partial images generated by dividing the image into a plurality of pieces while the image and the borders are superimposed together, an input unit that receives input of a label to be given to each of the plurality of partial images, and a generation unit that generates learning data for causing a learning model to learn by associating each of the plurality of partial images with the label given to the corresponding partial image of the plurality of partial images.